Machine Recognition of Devanagari Hand -printed Script...

Research Cell : An International Journal

ISSN: 2229-6913 (Print), ISSN: 2320-0332 (Online), Web Presence:

© 2016 Vidya Publications. Authors are responsible for any plagiarism issues.

Machine Recognition

Script: Character and Word Level Analysis

Department of Computer Science and Applications,

Panjab University

Abstract—Various South Asian languages such as Sanskrit, Hindi, Marathi, Nepali etc. are written using Devanagari script and

millions of people in world are using this script for written communication. So there is a high demand of good Devanagari ICR

still not available. An individual having good knowledge of the script of a language can easily read some words writte

pertaining to that script, though these are expressed in very bad manner, on the basis of his/her mental dictionary

be easy to read through a machine. A machine

upper, the lower and the middle. It is difficult to express the size of various regions in hand

bigger or smaller and depends upon a writer’s writing style.

the basis of structural features but in case

deals with the analysis of Devanagari characters and words according

structural features of Devanagari are also discussed.contoured image. The various problems likely to be faced

also discussed.

Keywords- Devanagari; hand-printed; structural, skeleton, contour,

1. Introduction

An ICR works in various stages such as scanning, pre

processing. Each stage has large number of optional techniques which have been used in different recognition

systems. Selecting a best technique for a particular application is a daunting task. One has to exhaustively study the

literature, implement them and observe t

method(s) used must be robust for expressing the properties of a character/script under consideration. If there is slight

variation in character image either due to printing, writing or

same. There are many feature extraction methods available in literature. The features may be local or global. The

features may be extracted from an original character or from its skeleton or contour

stage is concerned, Govindan et al[1] classified the various features in three categories

global transforms and series expansion. Each category has its pros and cons in terms of computatio

computational complicacy and accuracy. In structural based approach, a character is recognized on the basis of

structural primitives from which it is build up and these primitives are also known as character strokes. The number of

such stroke components, the kinds of stroke components and the relationship between these components in some order

is worked out. Some prominent authors have made prompt efforts to use structural features

extraction process from structural feature

Statistical Features are based on the statistical distribution of black and white pixels in a character image. These

features are invariants to character distortion and writing styl

Satish Kumar

Research Cell : An International Journal of Engineering Sciences, Issue June 2016, Vol. 18

0332 (Online), Web Presence: http://www.ijoes.vidyapublications.com

Authors are responsible for any plagiarism issues.

Machine Recognition of Devanagari Hand

: Character and Word Level Analysis

Satish Kumar

Department of Computer Science and Applications,

Panjab University SSG Regional Centre,

Hoshiarpur, Punjab, India

Email: [email protected]

Various South Asian languages such as Sanskrit, Hindi, Marathi, Nepali etc. are written using Devanagari script and

millions of people in world are using this script for written communication. So there is a high demand of good Devanagari ICR

An individual having good knowledge of the script of a language can easily read some words writte

script, though these are expressed in very bad manner, on the basis of his/her mental dictionary

machine-printed word in Devanagari script can be divided into three vertical regions

It is difficult to express the size of various regions in hand-printed i.e. the size

bigger or smaller and depends upon a writer’s writing style. In case of machine-printed it is easy to recognize a script on

the basis of structural features but in case of hand-printed it is difficult due to structural variation in writ

characters and words according to recognition point of view. The various issues related to the

structural features of Devanagari are also discussed. Moreover, the structural features are extracted from skeltonized or

problems likely to be faced while extracting structural features from skeltonized

structural, skeleton, contour, character recognition.

An ICR works in various stages such as scanning, pre-processing, feature extraction, classification and post

Each stage has large number of optional techniques which have been used in different recognition


literature, implement them and observe their performance. Obviously this is a big task.


variation in character image either due to printing, writing or due to instrument used, it should be able to absorb the


features may be extracted from an original character or from its skeleton or contour form. As far as feature extraction

stage is concerned, Govindan et al[1] classified the various features in three categories i.e.

global transforms and series expansion. Each category has its pros and cons in terms of computatio



mponents, the kinds of stroke components and the relationship between these components in some order

Some prominent authors have made prompt efforts to use structural features

is fast [1,2]. Section C covers drawbacks of using structural features.


features are invariants to character distortion and writing styles to some extent. In addition to this the feature vector

178

of Devanagari Hand-printed

: Character and Word Level Analysis

Various South Asian languages such as Sanskrit, Hindi, Marathi, Nepali etc. are written using Devanagari script and

millions of people in world are using this script for written communication. So there is a high demand of good Devanagari ICR but it is

An individual having good knowledge of the script of a language can easily read some words written on a paper

script, though these are expressed in very bad manner, on the basis of his/her mental dictionary. Such words may not

in Devanagari script can be divided into three vertical regions i.e. the

the size of a region can be

printed it is easy to recognize a script on

printed it is difficult due to structural variation in writing. This paper

The various issues related to the

Moreover, the structural features are extracted from skeltonized or

skeltonized or contoured image are

processing, feature extraction, classification and post-

Each stage has large number of optional techniques which have been used in different recognition


heir performance. Obviously this is a big task. The feature extraction


due to instrument used, it should be able to absorb the


As far as feature extraction

statistical, structural and

global transforms and series expansion. Each category has its pros and cons in terms of computational speed,



mponents, the kinds of stroke components and the relationship between these components in some order

Some prominent authors have made prompt efforts to use structural features[6,7,8,13]. The feature

covers drawbacks of using structural features.


es to some extent. In addition to this the feature vector




can be computed using these techniques with high speed.

recognition are zoning, moments, projection histograms, crossings, character loci a

methods based on profiles, crossings, histograms, zoning have been mostly used as complementary or supporting

features to enhance the performance of primary features [14

combination of statistical and structural features for handwritten character recognition[3, 5, 9,14].

above said feature extraction methods, some more features are also available which have very good dis

ability. Some of these features give very good results and are based on gradient (

chain code histograms of image contour[26

image (binary or gray)[34-35], stroke based[33

character[32], directional distance distribution[29

Segmentation is one among the various pre

classification is carried out. The variability caused in

even to read a hand-printed material by a well knower of the language and so reading the same script through a

machine cannot be expected. In automatic script recognition, a machine is trained to recognize the scripts of a

particular language so that it can decide the meaning of a

known as machine learning. In machine recognition process the same strategy is followed. Therefore, there are three

ways to automate a system for word recognition: 1) Segmentation Based, 2) Hol

based approach, the given word is segregated into individual components and each component is recognized and

assigned a symbol, and the resulting symbols are reassembled to know the identity of a word. Whereas in holistic

approach, a word is not broken down into individual components rather it is trained with a complete set of words by

extracting their properties. Some segmentation techniques used in Indian languages are given in [

The various efforts done for the recognition of Devanagari are reported by

who done work for Devanagari script recognition prior to 1990 are

script of Devanagari after 1990 are:[23-26]. The va

Sharma et al[54], Deshpande et al [55] and Kumar

recognition are: Shaw et al[58], Parui et al[59

Hindi is widely used language in south Asian region which is written in Devanagari script. Hindi is also official

language of India. This research paper covers the various issues related to the recognition of

characters and words. Section 2 covers Devanagari script analysis. Section 3 covers various structural issues in

character recognition. Section 4 covers various structural issues in word recognition.

complexities of Devanagari Words. Section

discussion and conclusion is coved in Section 7.

2. Script Analysis

In case of Devanagari script based languages, the words are written lineother majority of languages of world. So segmenting a page into lines and then lines into words follow the same strategy as it is used for other languages. language-dependent. Among Hindi and Sanskrit which are written using Devanagari script, it is difficult to recognize Sanskrit as its words are longer and compodifficult to locate not only in hand-printed but in cursive manner but there are some alphabets which writing decides whether its expression is cursive or not.

Satish Kumar




can be computed using these techniques with high speed. Some most commonly used statistical features for character

recognition are zoning, moments, projection histograms, crossings, character loci and n-tuple.


formance of primary features [14,16-18]. Some authors also have also used a

combination of statistical and structural features for handwritten character recognition[3, 5, 9,14].

above said feature extraction methods, some more features are also available which have very good dis

ability. Some of these features give very good results and are based on gradient ( Sobel or Prewit or Kirsch)[19

tograms of image contour[26-27], distance transform[28-29], pixel distance[30], a fixed size normalized

35], stroke based[33], feature based on foreground and backgroun

tional distance distribution[29] and directional element feature[31].

Segmentation is one among the various pre-processing steps performed on an image before

carried out. The variability caused in script writing is so high that sometimes

material by a well knower of the language and so reading the same script through a

In automatic script recognition, a machine is trained to recognize the scripts of a

particular language so that it can decide the meaning of a script under study. The process of training a machine is also


ways to automate a system for word recognition: 1) Segmentation Based, 2) Holistic, 3). Hybrid. In segmentation




Some segmentation techniques used in Indian languages are given in [

recognition of Devanagari are reported by Ghosh[64] and Pal [6

Devanagari script recognition prior to 1990 are :[47,50-52]. The major studies on machine

26]. The various studies on Devanagari character recognition are

and Kumar [56-57],. The various studies on hand-printed

Parui et al[59], Ramteke et al[61], Jayadevan et al[62], Oval[63]


This research paper covers the various issues related to the recognition of

and words. Section 2 covers Devanagari script analysis. Section 3 covers various structural issues in

character recognition. Section 4 covers various structural issues in word recognition. Section

Section 6 covers issues arising due to skeletonization of

discussion and conclusion is coved in Section 7.

In case of Devanagari script based languages, the words are written line-by line from top to bottom and left to right like languages of world. So segmenting a page into lines and then lines into words follow the same d for other languages. But segmentation technique required to segment a word into characters is

dependent. Among Hindi and Sanskrit which are written using Devanagari script, it is difficult to recognize Sanskrit as its words are longer and composed of a large number of half characters and / or vowels (matra) which are

printed but in machine printed too. The words in Devanagari script are not written in cursive manner but there are some alphabets which are written in cursive manner. Further, the style of individual writing decides whether its expression is cursive or not. In machine printed, variation in representation depends upon

179

Some most commonly used statistical features for character

tuple. The feature extraction


Some authors also have also used a

combination of statistical and structural features for handwritten character recognition[3, 5, 9,14]. In addition to

above said feature extraction methods, some more features are also available which have very good discrimination

Sobel or Prewit or Kirsch)[19-25] ,

], a fixed size normalized

], feature based on foreground and background information of a

erformed on an image before feature extraction and

sometimes it becomes difficult

material by a well knower of the language and so reading the same script through a

In automatic script recognition, a machine is trained to recognize the scripts of a

script under study. The process of training a machine is also


istic, 3). Hybrid. In segmentation




Some segmentation techniques used in Indian languages are given in [36-45].

Pal [65]. The various authors

]. The major studies on machine-printed

rious studies on Devanagari character recognition are: Pal et al[53],

printed Devanagari word

3] and Singh[60].


This research paper covers the various issues related to the recognition of Devanagari hand-printed

and words. Section 2 covers Devanagari script analysis. Section 3 covers various structural issues in

Section 5 covers recognition

of word level images. The

by line from top to bottom and left to right like languages of world. So segmenting a page into lines and then lines into words follow the same

But segmentation technique required to segment a word into characters is dependent. Among Hindi and Sanskrit which are written using Devanagari script, it is difficult to recognize

sed of a large number of half characters and / or vowels (matra) which are The words in Devanagari script are not written

Further, the style of individual In machine printed, variation in representation depends upon




the type-set of a font but in case of hand-print there is no limit.along their position in a word is given in T

Table 1: Devanagari Alphabet set along with their position in a word.

Like other languages, in Devanagari script too, a word is vertically divided into three regions: the upper region, the

middle region and the lower region. The upper region is occupied by the vowels or the part of vowels; the lower

region is also occupied by the vowels while the

compound characters, and a vowel (matra

three vertical regions is given in Fig 1.

Fig 1: Devanagari word divided into three vertical regions.

Some symbols mentioned as “Non Touching with Head

where as in some cases lower part of middle region is a little bit empty. The above said fact

machine–printed. Some full size, head

words in machine-printed form are depicted

Fig 2: The alphabets of various

3. Structural Issues in Character Recognition

Region

Upper

Middle full

Touching with

(Occupy upper part only)

Non-Touching with Head

Lower

Satish Kumar




print there is no limit. The basic Devanagari alphabetis given in Table 1.


n Devanagari script too, a word is vertically divided into three regions: the upper region, the


region is also occupied by the vowels while the middle region is occupied by the vowels, consonants, pure consonants,

matra) or a part of matra. A hand-printed Devanagari word demonstrating all the

Devanagari word divided into three vertical regions.

Touching with Head-line” in Table 1 cover whole lower part of middle region

where as in some cases lower part of middle region is a little bit empty. The above said fact

ome full size, head-line touching and non head-line touching characters in some Devanagari

printed form are depicted in Fig 2.

The alphabets of various categories Devanagari words.

in Character Recognition

Symbols

Touching with Head-line

(Occupy upper part only)

Touching with Head-line

180

Devanagari alphabets generally used in Hindi


n Devanagari script too, a word is vertically divided into three regions: the upper region, the


middle region is occupied by the vowels, consonants, pure consonants,

printed Devanagari word demonstrating all the

line” in Table 1 cover whole lower part of middle region

where as in some cases lower part of middle region is a little bit empty. The above said fact is true only for the

line touching characters in some Devanagari




Structural features are based on the geometrical and topological properties of local or global [1]. A character is composed of number of components in the form of strokes. lines, arcs, curves, etc. Each character component is called as a character primitive. These are extracted either from skeleton or contour of a character image. The various stroke primitives are extracted and approximated. The relationship between these components is established. The character is recognized on the basis of number of such components, the kinds of components and the relationship between these components in some order. The various structural methods differ in respect of primitive selection and their association for shape depiction. Some commonly used topological and geometrical features in character recognition are: endpoints, Tbottom, left and right), direction of stroke, loops, convepoints, etc.

3.1 Drawbacks of Structural Features for Characters Recognition:

approaches [5, 16] and these are as:

1). Since the number of primitives present in a character image is not known prior, it is very difficult to detect the

primitive and estimate its features. Therefore, the success of applying structural feature depends upon the prior

knowledge of shape boundary features stored in a databas

2). The matching schemes used in structural approaches use non

guarantee best match.

3). The structural features are sensitive to noise and this representation do not preserve the topological structure of

object. With change in boundary of an image the local features change a lot.

4). In structural based approach, though the character primitives provide a stable representation but it does not

completely cover the variability in characters. The

variable writing styles is not easy to handle in classification stage.

Moreover, the structural features are extracted from a skeleton. Thinning process introduces spurious branches and

clusters as given in Fig 3. These defects are common in handwritten character images. The spurious branches and

clusters of small size are easy to remove but bigger size poses difficulty.

Fig 3: Spurious branches and clusters created due to thinning process.

Even performing primary classification based on topological features in Devanagari handwritten character recognition

is difficult. For example, if we want to categorize Devanagari alphabet set into subsets on the basis of number of end

points. It is very difficult as the number of end points of a character is different corresponding to different writing. The

various intra-class Devanagari handwritten characters with varying number of en

characters are taken after head-line removal.

Satish Kumar




Structural features are based on the geometrical and topological properties of a character and these prop]. A character is composed of number of components in the form of strokes.

lines, arcs, curves, etc. Each character component is called as a character primitive. These are extracted either from skeleton or contour of a character image. The various stroke primitives are extracted and approximated. The

ship between these components is established. The character is recognized on the basis of number of such components, the kinds of components and the relationship between these components in some order. The various

rimitive selection and their association for shape depiction. Some commonly used topological and geometrical features in character recognition are: endpoints, T-points, cross points, extrema (top, bottom, left and right), direction of stroke, loops, convex and concave arcs, straight lines, directi

Drawbacks of Structural Features for Characters Recognition: There are some drawbacks of

s present in a character image is not known prior, it is very difficult to detect the


knowledge of shape boundary features stored in a database.

he matching schemes used in structural approaches use non-metric similarity measure. These methods do not

he structural features are sensitive to noise and this representation do not preserve the topological structure of

object. With change in boundary of an image the local features change a lot.

n structural based approach, though the character primitives provide a stable representation but it does not

completely cover the variability in characters. The instability caused in feature space due to incomplete recovery of

variable writing styles is not easy to handle in classification stage.


. These defects are common in handwritten character images. The spurious branches and

clusters of small size are easy to remove but bigger size poses difficulty.

: Spurious branches and clusters created due to thinning process.



difficult as the number of end points of a character is different corresponding to different writing. The

class Devanagari handwritten characters with varying number of end points are given in Fig 4

line removal.

181

and these properties may be ]. A character is composed of number of components in the form of strokes. These strokes may be

lines, arcs, curves, etc. Each character component is called as a character primitive. These are extracted either from skeleton or contour of a character image. The various stroke primitives are extracted and approximated. The

ship between these components is established. The character is recognized on the basis of number of such components, the kinds of components and the relationship between these components in some order. The various

rimitive selection and their association for shape depiction. Some commonly points, cross points, extrema (top,

x and concave arcs, straight lines, directional points, bend

There are some drawbacks of using structural

s present in a character image is not known prior, it is very difficult to detect the


metric similarity measure. These methods do not

he structural features are sensitive to noise and this representation do not preserve the topological structure of an

n structural based approach, though the character primitives provide a stable representation but it does not

instability caused in feature space due to incomplete recovery of


. These defects are common in handwritten character images. The spurious branches and

: Spurious branches and clusters created due to thinning process.



difficult as the number of end points of a character is different corresponding to different writing. The

d points are given in Fig 4. All these




Fig 4: Some Devanagari handwritten characters with varying number of end points.

Some more reasons for not using structural features for Devanagari handwritten character recognition are as

follows[46]:

1). Extracting structural features are very difficult

a) Due to complex structure of some of its characters.

b) Intra-touching (mingle) of the various strokes present in a character

style of writing or ink blot.

Fig 5: Some Devanagari characters with

c) Some characters are built up from multi

character, Fig 6, due to hasty writing or individual style of writing.

Fig 6: Multi

d) There is blurring present in some characters. Extracting skeleton from blurred image destroy its shape

demonstrates the situation, where

Fig 7: Blurred

Satish Kumar




: Some Devanagari handwritten characters with varying number of end points.


are very difficult

) Due to complex structure of some of its characters.

various strokes present in a character, Fig 5, due to hasty

: Some Devanagari characters with intra-touching strokes.

) Some characters are built up from multi-strokes and these primitive may not be touching as it is required in a

due to hasty writing or individual style of writing.

ulti-stroke characters with dissociated primitives.

There is blurring present in some characters. Extracting skeleton from blurred image destroy its shape

, where encircled parts of the characters are completely vanished

: Blurred (encircled) and vanished Devanagari characters

182

: Some Devanagari handwritten characters with varying number of end points.


hasty writing or individual

strokes and these primitive may not be touching as it is required in a

There is blurring present in some characters. Extracting skeleton from blurred image destroy its shape. Fig 7

vanished.




2). It is difficult to extract structural features from noisy images as small loops are created in skeletonization process.

3). Extra branches are created due to fluent

skeletonized image of a character. These extra branches become more apparent after skeltonization of a character

image is carried. Such unpredictable behavior is common

Fig

4. Structural Issues in Word Recognition

In segmentation based approach, a given word is segregated into individual components and each component is

recognized and assigned a symbol, and the resulting symbols are reassembled to know the identity of a word.

printed writing the variability caused is so high that

regions as the size of different characters varies due to the lack of control while writing. The

demonstrates this situation. As a head-line exists in case of words / characters written in D

to locate boundary between upper and middle regions and helpful in segregating /

region. It is very difficult to locate the boundary between

Fig 9: Some full size, head-line touching and non

Though some characters of Devanagari script are cursive in writing but while writing a word each character is written

individually. A Devanagari word is formed by connecting its letters with each other through head

character in a word may touch with adjoining character either due to hasty writing or due to ink stains.

Fig 10: A Hindi word

In case of Devanagari, some vowels are formed with the symbols those occupy both upper and middle

recognition process, it is important to decide about a vowel by considering the various parts of a symbol presented

both in upper and lower regions which are some what difficult to locate in the sense that they may be written / placed

in a word deviating from the rules. The recognition process may produce the output of such image as ambiguous word.

The possible word images with exact location of vowels for the word image demonstrated in

with minor change in location of symbols in upper or middle creates a new word as given in Fig 10(right).

Satish Kumar





due to fluent writing as shown in Fig 8. Structural features are extracted from the

These extra branches become more apparent after skeltonization of a character

Such unpredictable behavior is common in skeletonizing algorithms [66].

8: Extra branches produced due to writing.

Recognition

given word is segregated into individual components and each component is

recognized and assigned a symbol, and the resulting symbols are reassembled to know the identity of a word.

is so high that it is very difficult to locate boundary between middle and lower

regions as the size of different characters varies due to the lack of control while writing. The

line exists in case of words / characters written in Devanagari script, it is

er and middle regions and helpful in segregating / locatin

region. It is very difficult to locate the boundary between middle and lower region.

line touching and non head-line touching characters in some Devanagari words in

hand-printed form.


A Devanagari word is formed by connecting its letters with each other through head

touch with adjoining character either due to hasty writing or due to ink stains.

word where matra not in its original position(right).

In case of Devanagari, some vowels are formed with the symbols those occupy both upper and middle



The recognition process may produce the output of such image as ambiguous word.

The possible word images with exact location of vowels for the word image demonstrated in

mbols in upper or middle creates a new word as given in Fig 10(right).

183


. Structural features are extracted from the

These extra branches become more apparent after skeltonization of a character

].

given word is segregated into individual components and each component is

recognized and assigned a symbol, and the resulting symbols are reassembled to know the identity of a word. In hand-

t is very difficult to locate boundary between middle and lower

regions as the size of different characters varies due to the lack of control while writing. The Fig 9 clearly

evanagari script, it is easy

locating the symbols of upper

line touching characters in some Devanagari words in


A Devanagari word is formed by connecting its letters with each other through head-line. However, a

touch with adjoining character either due to hasty writing or due to ink stains.

.

In case of Devanagari, some vowels are formed with the symbols those occupy both upper and middle regions. In



The recognition process may produce the output of such image as ambiguous word.

The possible word images with exact location of vowels for the word image demonstrated in Fig 10(left). However,

mbols in upper or middle creates a new word as given in Fig 10(right).




In addition to this, some symbols of Devanagari script are built up with multiple strokes which may be dissociated

from each other in a given word. In such cases decision about a chara

available stroke components. In Fig 11

difficulty in segmenting and recognizing such characters.

Fig 11: A word written in Devanagari script having broken character primitives

Among the various middle region full size symbols only ‘

vowel and remaining symbols are either consonants or pure consona

middle region, Fig 12, may create confusion with noise present in an image.

frequently used vowel in Devanagari wher

Fig 12: A matra or

In Hindi it is scarcely used and in Sanskrit it

demonstrated in Fig 12. The use of visarga is very less in middle of a word where its use at end of word is common

in Sanskrit. Like Roman, “-” (hyphen) is also used in Devanagari to join two words. This also lies in middle region

but its presence is outside a word. For recognition purpose, it also needs to be treated as

5. Recognition Complexities of Devanagari Words

There are three kinds of approaches to recognize the words of any language

(holistic) and hybrid. The segmentation free approach, as already mentioned, requires a big lexicon of words which

needs a lot of efforts for lexicon creation. The segmentation

segmentation process in it. Some issues required to be dealt with the recognition of words pertaining to Devanagari

scripts using this approach are as follows:

1) There are two ways to segregate Devanagari words into characters and then recognize them individually

either after removing head-line or without removing head

box characters\matra chopped off due to which it poses a problem to recognize

present in a word. How the head-line f

various characters in a given word image?

2) If a word is segmented without removing head

character in a word which is non-linear.

characters even if two adjoining character

Satish Kumar





from each other in a given word. In such cases decision about a character must be made keeping in view the various

ble stroke components. In Fig 11, are in broken form and there may be

difficulty in segmenting and recognizing such characters.

A word written in Devanagari script having broken character primitives

Among the various middle region full size symbols only ‘ ’ (kanna) and ‘:’ (visarga) belong to vowel or a part of

vowel and remaining symbols are either consonants or pure consonants or compound characters.

may create confusion with noise present in an image. The symbol

frequently used vowel in Devanagari where as the use of symbol ‘:’ (visarg) is language dependen

A matra or any other symbol in middle region.

In Hindi it is scarcely used and in Sanskrit it is frequently used. The use of ‘ ’ and ‘:’ in middle

The use of visarga is very less in middle of a word where its use at end of word is common

” (hyphen) is also used in Devanagari to join two words. This also lies in middle region

For recognition purpose, it also needs to be treated as a special symbol.

Recognition Complexities of Devanagari Words

There are three kinds of approaches to recognize the words of any language i.e. segmentation based, segmentation free

The segmentation free approach, as already mentioned, requires a big lexicon of words which

needs a lot of efforts for lexicon creation. The segmentation based approach is easy but it is

ssues required to be dealt with the recognition of words pertaining to Devanagari

as follows:

There are two ways to segregate Devanagari words into characters and then recognize them individually

line or without removing head-line. While removing head-line some parts of middle

chopped off due to which it poses a problem to recognize various individual

line from a word may be removed without chopping off important parts of

various characters in a given word image?

If a word is segmented without removing head-line, it is difficult to estimate the left and right boundary of a

near. Also it becomes difficult to decide the dissociati

characters are not fused.

184


cter must be made keeping in view the various

are in broken form and there may be

A word written in Devanagari script having broken character primitives

) belong to vowel or a part of

nts or compound characters. Visarga lies in

The symbol ‘ ’ (kanna ) is most

rg) is language dependent.

’ and ‘:’ in middle region is

The use of visarga is very less in middle of a word where its use at end of word is common

” (hyphen) is also used in Devanagari to join two words. This also lies in middle region

special symbol.

segmentation based, segmentation free

The segmentation free approach, as already mentioned, requires a big lexicon of words which

it is difficult to carry out

ssues required to be dealt with the recognition of words pertaining to Devanagari

There are two ways to segregate Devanagari words into characters and then recognize them individually i.e.

line some parts of middle

various individual characters

rom a word may be removed without chopping off important parts of

line, it is difficult to estimate the left and right boundary of a

Also it becomes difficult to decide the dissociation point between two




3) Due to hasty writing, the two adjoining characters may

touching characters?

4) Some multi-stroke characters, Fig 1

those components are recognized to take decision about the exact/ tentative cha

5) There may be a vowel attached to a character on its lower part.

the presence/ absence of a vowel on the lower part of a character?

6) There are some vowels in Devanagari,

regions. How to locate and recognize such vowels?

7) Some pure consonants (half characters) presented inside a word may form a complex structure by combining

with other characters. Some such complex structures formed due to the merging of a consonant and a pure

consonant are demonstrated inside dotted

components?

Fig 13: Some complex structures formed due to the merging o

8) Various matra attached on lower or upper part of a character may be bigger in size due to which it encroach

the areas of the adjoining character

equivalent to characters size to which it is attached.

aggregates the ambiguity about final word recognition.

9) Various complex compound characters are formed due to merge of two or more characters.

characters are written left to right, the fusion is expected

but this is not only the case in

inside dotted rectangle, making it difficult to know whether to dissociate

horizontally or vertically.

Fig 14 (

Satish Kumar




two adjoining characters may touch each other. How to locate the boundary between

, Fig 11, in a word may not be connected to each other as these

those components are recognized to take decision about the exact/ tentative character?

There may be a vowel attached to a character on its lower part. Fig 9 demonstrates the situation.

the presence/ absence of a vowel on the lower part of a character?

There are some vowels in Devanagari, Fig 10(right), which may be mis-positioned both in upper

regions. How to locate and recognize such vowels?

Some pure consonants (half characters) presented inside a word may form a complex structure by combining

Some such complex structures formed due to the merging of a consonant and a pure

consonant are demonstrated inside dotted rectangle areas in Fig 13. How to locate and recognize such

Some complex structures formed due to the merging of consonants and pure consonants.

attached on lower or upper part of a character may be bigger in size due to which it encroach

the areas of the adjoining characters. In machine printed the size of such matra or a part of matra

equivalent to characters size to which it is attached. Such matra may create confusion

aggregates the ambiguity about final word recognition. Fig 14 depicts the situation.

Fig 14: Confused location of lower Matra.

Various complex compound characters are formed due to merge of two or more characters.

ten left to right, the fusion is expected horizontal as given in Fig 14(a), inside dotted circle,

in Devanagari where the fusion also occurs vertically as given in Fig 14(a

making it difficult to know whether to dissociate such fused

Fig 14 (a): Horizontal and vertically fused characters.

185

each other. How to locate the boundary between

connected to each other as these should be. How

the situation. How to decide

positioned both in upper or middle

Some pure consonants (half characters) presented inside a word may form a complex structure by combining

Some such complex structures formed due to the merging of a consonant and a pure

How to locate and recognize such

f consonants and pure consonants.

attached on lower or upper part of a character may be bigger in size due to which it encroach

the size of such matra or a part of matra remains

confusion about its location. This

Various complex compound characters are formed due to merge of two or more characters. Since, the

horizontal as given in Fig 14(a), inside dotted circle,

vertically as given in Fig 14(a) ,

such fused compound characters




10) Some compound characters, built up due to combine of two

symbol in Devanagari as given in Fig 14(b).

complexity for recognition. The numbers of such characters

Fig 14 (

11) Some awkward styles of writing such as undeveloped, unusual and extra

Devanagari word recognition. In addition some irregularities committed by the writers while writing further

adds complicacies in Devanagari word recognition problem

6. Word Level Skeletonization or Contouring

A word is composed of number of characters each of which is formed from a number of stroke primitives. These

strokes may be lines, arcs, curves, etc and may or may not be connected to each other depending upon the structure

of a character. The stroke primitives can be extracted from eith

structural based recognition process, the various stroke primitives of a character are extracted and approximated.

The relationships between various stroke components are established. Some difficulties in rec

pertaining to Devanagari script using structural approach

word are as follows:

1). A Devanagari word consists of a head

produced is not smooth which is difficult to remove. One such

given in Fig 15 mentioned using dotted

2). When a word is segmented after skeletonization, the two characters which are fused may form a cluster or

multiple joint which is difficult to segment. The situation is given in

3). In segmentation based approach, each segmented character/ part of character is matched against some already

known characters. If a character is in skeleton form, then the skeletonization process may have lost important

part of a character image or may have created extra

character image etc., which complicates the recognition process a lot.

Fig 15: Skeletonized image of hand

4). The various character level recognition 3.1 further adds to word level recognition

Satish Kumar




ome compound characters, built up due to combine of two characters, are also represented using special

symbol in Devanagari as given in Fig 14(b). These symbols form a complex structure and

The numbers of such characters are more in Sanskrit than Hindi.

Fig 14 (b): Special symbol for fused characters.

Some awkward styles of writing such as undeveloped, unusual and extra stroke writing

In addition some irregularities committed by the writers while writing further

Devanagari word recognition problem[67].

or Contouring Issues

characters each of which is formed from a number of stroke primitives. These


of a character. The stroke primitives can be extracted from either skeleton or contour of a character image. In


The relationships between various stroke components are established. Some difficulties in rec

pertaining to Devanagari script using structural approach which is based on skeletonized

A Devanagari word consists of a head-line. When the given word image is skeletonized, the resultant head

produced is not smooth which is difficult to remove. One such irregular structure produced in

mentioned using dotted rectangle area.

When a word is segmented after skeletonization, the two characters which are fused may form a cluster or

multiple joint which is difficult to segment. The situation is given in Fig 15 mentioned using dotted circles.

based approach, each segmented character/ part of character is matched against some already


part of a character image or may have created extra parts in a character image or may have created clusters in a

character image etc., which complicates the recognition process a lot.

Skeletonized image of hand-printed Devanagari word image

The various character level recognition problems based on skeletonized character images mentioned in Section recognition.

186

represented using special

a complex structure and are a source of

more in Sanskrit than Hindi.

stroke writing also pose problems in

In addition some irregularities committed by the writers while writing further

characters each of which is formed from a number of stroke primitives. These


er skeleton or contour of a character image. In


The relationships between various stroke components are established. Some difficulties in recognizing words

skeletonized or contoured image of a

line. When the given word image is skeletonized, the resultant head-line

irregular structure produced in head-line is

When a word is segmented after skeletonization, the two characters which are fused may form a cluster or

mentioned using dotted circles.

based approach, each segmented character/ part of character is matched against some already


parts in a character image or may have created clusters in a

ri word image.

er images mentioned in Section




5). During recognition process, the character images are normalized so that proper mapping between stored image and tested images is done. But itnormalization.

6). On the other hand, if a character\word structure is compared from the problems. The contoured image of a wordboundary where extra stokes of a character are generated

Fig 16: The co

7). Some unwanted complex structures are formarea, due to which segmentation or comparison becomes difficult.

Fig 17: The co

8). Moreover, if character\word strokes do not contain adequate size due to poor scanning, thin writing instrument tip, resizing or other reason, the contoured image of such characterresultant image becomes the mixer of contour and skeleton.

Fig 18: The skeleton strokes produced, dotted circles, during contouring.

Satish Kumar




recognition process, the character images are normalized so that proper mapping between stored image But it is difficult to decide that whether first skeletonization is performe

word structure is compared from the contoured imagetoured image of a word, Fig 9, is given in Fig 16. An unwanted structure (lump

boundary where extra stokes of a character are generated or ink blot occurs on boundary during writing process.

The contoured image of a word given in Fig 9.

complex structures are formed on fusion point of two or more characters, Figsegmentation or comparison becomes difficult.

The contoured image of a word given in Fig 13.

kes do not contain adequate size due to poor scanning, thin writing instrument tip, resizing or other reason, the contoured image of such character\word produce a skeleton as given in Fig 18. resultant image becomes the mixer of contour and skeleton. This also poses difficulty in comparison.

The skeleton strokes produced, dotted circles, during contouring.

187

recognition process, the character images are normalized so that proper mapping between stored image is difficult to decide that whether first skeletonization is performed or

contoured image then it has its own An unwanted structure (lump) is create on

during writing process.

of two or more characters, Fig 17 doted circular

kes do not contain adequate size due to poor scanning, thin writing instrument tip, word produce a skeleton as given in Fig 18. The

This also poses difficulty in comparison.

The skeleton strokes produced, dotted circles, during contouring.




9). The images resized with normalization process also, sometimes, destroys strokes present in a character/word image due to which it becomes difficult to extract contour.

Some above mentioned issues are not only scripts.

7. Discussion and Conclusion

Though a lot of work has been done for the development of recognition system for the various languages of the world

but good ICR of majority of languages are still not available. Among the various Indian scri

script. The words of Devanagari script are formed by connecting various characters using head

presence of head-line it is easy to separate the symbols

becomes difficult to separate the symbols between

boundary. Though its words are not cursively written but some its characters are cursive i

has its own kinds of problems in respect of recognition. We have

of Devanagari hand-printed based on structural features.

structural features which are extracted from skeltonized

related to structural features also discussed.

discussed.

References

[1] V. K. Govindan and A. P. Shivaprasad, “Character Recognition

[2] M. S. Khorsheed, “Off-line Arabic Character Recognition

45 (2002).

[3] H. S. Baird, “Feature Identification for Hybrid Structural/Statistical Pattern Classification”, Computer Vision, Graphics and

Image Processing, Vol. 42, pp. 318-333(1988).

[4] K. Anisimovich, V. Rybkin, A. Shamis and V.

Classifiers for Recognition of Hand-printed Characters”, Proceedings of Fourth International Conference of Document

Analysis and Recognition, Ulm, Germany, pp. 881

[5] P. Foggia, C. Sansone, F. Tortorella and M. Vento, “Combining Statistical and Structural Approaches for Handwritten

Character Description”, Image and Vision Computing, Vol. 17, pp. 701

[6] X. Li, W. Oh, J. Hong and W. Gao, “Recognizing Component

with Stable Features”, Proceedings of the International Conference on Document Analysis and Recognition, Ulm,

Germany, pp. 616-620(1997).

[7] J. Rocha and T. Puvlidis, “A Shape Analysis Model wi

Transactions on Pattern Analysis and Machine Intelligence

[8] K. T. Miura, R. Sato and S. Mori, “A Method of Extracting Curvature Features and Its Applications

Character Recognition”, Proceedings of the International Conference on Document Analysis and Recognition, Ulm,

Germany, pp. 450-454(1997).

[9] J. Cai and Z-Q Liu, “Integration of Structural

Recognition,” IEEE Transactions on Pattern Analysis and Machine

[10] G. S. Lehal and Chandan Singh, “

Recognition, Barcelona, Spain, Vol. 2, pp.

Satish Kumar




9). The images resized with normalization process also, sometimes, destroys strokes present in a character/word image due to which it becomes difficult to extract contour.

not only hindrance in Devanagari based script recognition but


languages are still not available. Among the various Indian scri

The words of Devanagari script are formed by connecting various characters using head

line it is easy to separate the symbols lying between upper region and middle region.

becomes difficult to separate the symbols between middle and lower region due to absence of any such line/

Though its words are not cursively written but some its characters are cursive in writing. Devanagari script

has its own kinds of problems in respect of recognition. We have studied the problems associated with the recognition

printed based on structural features. Also, it is difficult to recognize the hand

structural features which are extracted from skeltonized or contoured character/word images. The various issues

related to structural features also discussed. Some issues related to segmentation of hand-printed Devanagari

V. K. Govindan and A. P. Shivaprasad, “Character Recognition – a Review”, Pattern Recognition, Vol. 23, No. 7 (1990).

line Arabic Character Recognition - a Review”, Pattern Analysis and Applications, V

H. S. Baird, “Feature Identification for Hybrid Structural/Statistical Pattern Classification”, Computer Vision, Graphics and

333(1988).

K. Anisimovich, V. Rybkin, A. Shamis and V. Tereshchenko , “Using Combination of Structural, Feature and Raster

printed Characters”, Proceedings of Fourth International Conference of Document

Analysis and Recognition, Ulm, Germany, pp. 881-885 (1997).

oggia, C. Sansone, F. Tortorella and M. Vento, “Combining Statistical and Structural Approaches for Handwritten

Character Description”, Image and Vision Computing, Vol. 17, pp. 701-711(1999).

Recognizing Components of Handwritten Characters by Attributed Relational Graphs


J. Rocha and T. Puvlidis, “A Shape Analysis Model with Applications to a Character Recognition System”, IEEE

Transactions on Pattern Analysis and Machine Intelligence, Vol. 16, No. 4, pp. 393-405 (1994).

K. T. Miura, R. Sato and S. Mori, “A Method of Extracting Curvature Features and Its Applications


Integration of Structural and Statistical Information For Unconstrained

Pattern Analysis and Machine Intelligence, Vol. 21, pp. 263

, “A Gurmukhi Script Recognition System”, International

Spain, Vol. 2, pp. 557-560(2000).

188

9). The images resized with normalization process also, sometimes, destroys strokes present in a character/word image

hindrance in Devanagari based script recognition but are ubiquitous to all


languages are still not available. Among the various Indian scripts, Devanagari is such

The words of Devanagari script are formed by connecting various characters using head-line. Due to the

lying between upper region and middle region. However, it

due to absence of any such line/

n writing. Devanagari script

studied the problems associated with the recognition

Also, it is difficult to recognize the hand-printed script using

character/word images. The various issues

printed Devanagari are also

a Review”, Pattern Recognition, Vol. 23, No. 7 (1990).

a Review”, Pattern Analysis and Applications, Vol. 5, pp. 31-

H. S. Baird, “Feature Identification for Hybrid Structural/Statistical Pattern Classification”, Computer Vision, Graphics and

Tereshchenko , “Using Combination of Structural, Feature and Raster

printed Characters”, Proceedings of Fourth International Conference of Document

oggia, C. Sansone, F. Tortorella and M. Vento, “Combining Statistical and Structural Approaches for Handwritten

s of Handwritten Characters by Attributed Relational Graphs


th Applications to a Character Recognition System”, IEEE

K. T. Miura, R. Sato and S. Mori, “A Method of Extracting Curvature Features and Its Applications to Handwritten


Unconstrained Handwritten Numeral

Intelligence, Vol. 21, pp. 263-270(1999).

International Conference on Pattern




[11] B. B. Chaudhuri and U. Pal, “A Complete

(1998).

[12] R. C. Gonzalez and R. E. Woods, “Digital

[13] H. Xue and V.Govindaraju, “Building Skeletal Graphs for Structural Feature

Proceedings of the Sixth International

100(2001).

[14] L. Heutte, T. Paquet, J. Moreau, Y. Lecourtier and C. Olivier, “A Structural /

Handwritten Character Recognition”, Pattern Recognition Letter, Vol. 19, pp. 629

[15] D. Zhang and G. Lu, “Review of Shape Representation and Description Techniques”, Pattern Recognition, Vol. 37, pp. 1

19 (2004).

[16] N. Arica and F. T. Yarman-Vural, “Optical Character Recognition for Cursive Handwriting”, IEEE Transactions on Pattern

Analysis and Machine Intelligence, Vol. 24, No. 6(2002).

[17] K. M. Kim, J.J. Park, Y.G. Song, I. C. Kim and C. Y. Suen, “Recognition of Handwritten Numerals Using a Combined

Classifier with Hybrid Features”, SSPR & SPR, LNCS 3138, pp. 992

[18] J. T. Favata, G. Srikantan and S. N. Srihari, “Handprinted Character / Digit Recognition using a Multiple Feature/Resolution

Philosophy”, Proceedings of the International Workshop on Frontiers of Handwriting Recognition”, pp. 57

[19] K. Roy, T. Pal, U. Pal and F. Kimura, “Ori

International Conference on Document Analysis and Recognition, Vol. 2, pp. 770

[20] G. Srikantan, S. W. Lam and S.N. Srihari, “Gradient

Recognition, Vol. 29, No. 7, pp. 1147-

[21] H. Liu and X. Ding, “Handwritten Character Recognition using Gradient Feature and Quadratic Classifier with Multiple

Discrimination Schemes”, Proceedings of the Eighth

19-25(2005).

[22] H. Fujisawa and C.–L. Liu, “Directional Pattern Matching for Character Recognition Revisited”, Proceedings of the

Seventh International Conference on Document Analysis a

[23] S.-B. Cho, “Neural-Network Classifiers for Recognizing Totally Unconstrained Handwritten Numerals”, IEEE Transactions

on Neural Networks, Vol. 4, No. 1, pp. 43

[24] K. M. Kim, J.J. Park, Y.G. Song, I. C.

Classifier with Hybrid Features”, SSPR & SPR, LNCS 3138, pp. 992

[25] S. Knerr, L. Personnaz and G. Dreyfus, “Handwritten Digit Recognition by Neural Networks with Sin

IEEE Transactions on Neural Networks, Vol. 3, No. 6, pp. 962

[26] F. Kimura and M. Shridhar, Handwritten Numeral Recognition Based on Multiple Algorithms, Pattern Recognition, Vol.

24, No. 10, pp. 969-983(1991).

[27] J. Cao, M. Ahmadi and M. Shridhar, Recognition of Handwritten Numerals with Multiple Feature and Multistage Classifier,

Pattern Recognition, Vol. 28, No. 2, pp. 153

[28] Zs. M. Kovics and R. Guerrieri, “Massively

Pattern Recognition, Vol. 28, No. 3, pp. 293

[29] II-S. Oh, C.Y. Suen, “Distance Features for Neural Network

Journal on Document Analysis and Recognition, Vol.1, pp. 73

[30] N. W. Strathy and C. Y. Suen, “A New System for Reading Handwritten ZIP Codes”, Proceedings of the International

Conference on Document Analysis and Recognition, Montreal, Canada, pp. 74

Satish Kumar




A Complete Printed Bangla OCR System”, Pattern Recognition, Vol. 31, No. 5, pp. 531

Digital Image Processing”, 2nd Ed. 2002, Pearson Education.

H. Xue and V.Govindaraju, “Building Skeletal Graphs for Structural Feature Extraction on Handwriting Images”

Conference on Document Analysis and Recognition, Seattle, WA, USA, pp.

L. Heutte, T. Paquet, J. Moreau, Y. Lecourtier and C. Olivier, “A Structural / Statistical Feature Based V

Pattern Recognition Letter, Vol. 19, pp. 629-641(1998).

pe Representation and Description Techniques”, Pattern Recognition, Vol. 37, pp. 1

Vural, “Optical Character Recognition for Cursive Handwriting”, IEEE Transactions on Pattern

Vol. 24, No. 6(2002).

K. M. Kim, J.J. Park, Y.G. Song, I. C. Kim and C. Y. Suen, “Recognition of Handwritten Numerals Using a Combined

Classifier with Hybrid Features”, SSPR & SPR, LNCS 3138, pp. 992-1000(2004).

S. N. Srihari, “Handprinted Character / Digit Recognition using a Multiple Feature/Resolution

Philosophy”, Proceedings of the International Workshop on Frontiers of Handwriting Recognition”, pp. 57

] K. Roy, T. Pal, U. Pal and F. Kimura, “Oriya Handwritten Numeral Recognition System”, Proceedings of the Eight

International Conference on Document Analysis and Recognition, Vol. 2, pp. 770- 774 (2005).

] G. Srikantan, S. W. Lam and S.N. Srihari, “Gradient-Based Contour Encoding for Chara

-1160 (1996).

H. Liu and X. Ding, “Handwritten Character Recognition using Gradient Feature and Quadratic Classifier with Multiple

Discrimination Schemes”, Proceedings of the Eighth International Conference on Document Analysis and Recognition, pp.

L. Liu, “Directional Pattern Matching for Character Recognition Revisited”, Proceedings of the

Seventh International Conference on Document Analysis and Recognition, pp. 794-798(2003).

Network Classifiers for Recognizing Totally Unconstrained Handwritten Numerals”, IEEE Transactions

on Neural Networks, Vol. 4, No. 1, pp. 43-53(1997).

] K. M. Kim, J.J. Park, Y.G. Song, I. C. Kim and C. Y. Suen, “Recognition of Handwritten Numerals Using a Combined

Classifier with Hybrid Features”, SSPR & SPR, LNCS 3138, pp. 992-1000(2004).

S. Knerr, L. Personnaz and G. Dreyfus, “Handwritten Digit Recognition by Neural Networks with Sin

IEEE Transactions on Neural Networks, Vol. 3, No. 6, pp. 962-968(1992).

F. Kimura and M. Shridhar, Handwritten Numeral Recognition Based on Multiple Algorithms, Pattern Recognition, Vol.

Cao, M. Ahmadi and M. Shridhar, Recognition of Handwritten Numerals with Multiple Feature and Multistage Classifier,

Pattern Recognition, Vol. 28, No. 2, pp. 153-160(1995).

Zs. M. Kovics and R. Guerrieri, “Massively-Parallel Handwritten Character Recognition Based on the Distance Transform”,

Pattern Recognition, Vol. 28, No. 3, pp. 293-301(1995).

S. Oh, C.Y. Suen, “Distance Features for Neural Network-based Recognition of Handwritten Characters”, International

and Recognition, Vol.1, pp. 73-88(1998).

N. W. Strathy and C. Y. Suen, “A New System for Reading Handwritten ZIP Codes”, Proceedings of the International

Conference on Document Analysis and Recognition, Montreal, Canada, pp. 74-77(1995).

189

Pattern Recognition, Vol. 31, No. 5, pp. 531-549

rson Education.

Extraction on Handwriting Images”,

and Recognition, Seattle, WA, USA, pp. 96-

Statistical Feature Based Vector for

pe Representation and Description Techniques”, Pattern Recognition, Vol. 37, pp. 1-

Vural, “Optical Character Recognition for Cursive Handwriting”, IEEE Transactions on Pattern

K. M. Kim, J.J. Park, Y.G. Song, I. C. Kim and C. Y. Suen, “Recognition of Handwritten Numerals Using a Combined

S. N. Srihari, “Handprinted Character / Digit Recognition using a Multiple Feature/Resolution

Philosophy”, Proceedings of the International Workshop on Frontiers of Handwriting Recognition”, pp. 57-66(1994).

ya Handwritten Numeral Recognition System”, Proceedings of the Eight

Based Contour Encoding for Character Recognition”, Pattern

H. Liu and X. Ding, “Handwritten Character Recognition using Gradient Feature and Quadratic Classifier with Multiple

International Conference on Document Analysis and Recognition, pp.

L. Liu, “Directional Pattern Matching for Character Recognition Revisited”, Proceedings of the

Network Classifiers for Recognizing Totally Unconstrained Handwritten Numerals”, IEEE Transactions

Kim and C. Y. Suen, “Recognition of Handwritten Numerals Using a Combined

S. Knerr, L. Personnaz and G. Dreyfus, “Handwritten Digit Recognition by Neural Networks with Single-Layer Training”,

F. Kimura and M. Shridhar, Handwritten Numeral Recognition Based on Multiple Algorithms, Pattern Recognition, Vol.

Cao, M. Ahmadi and M. Shridhar, Recognition of Handwritten Numerals with Multiple Feature and Multistage Classifier,

ecognition Based on the Distance Transform”,

based Recognition of Handwritten Characters”, International

N. W. Strathy and C. Y. Suen, “A New System for Reading Handwritten ZIP Codes”, Proceedings of the International




[31] N. Kato, M. Suzuki, S. Omachi, H. Aso and Y. Nemoto, “A Handwritten Character Recognition System using Directional

Element Feature and Asymmetric Mahalanobis Distance”, IEEE Transactions on Pattern Recognition Analysis and machine

Intelligence, Vol. 21, No. 3(1999).

[32] S. Britto Jr., R. Sabourin, F. Bortolozzi and C. Y. Suen, “Foreground and Background Information in an HMM

Method for Recognition of Isolated Characters and Numeral Strings”, Ninth International Workshop on Frontiers in

Handwriting Recognition, Kokubunji, Tokyo, Japan, pp. 371

[33] T. W. Bhowmik, U. Bhattacharya and S.K. Parui, “Recognition of Bangla Handwritten Characters using an MLP Classifier

Based on Stroke Features”, Proceedings of ICONIP,

[34] K. Roy, S. Vajda, U. Pal and B.B. Chaudhuri, “ A System Towards Indian Postal Automation”, Proceedings of the Ninth

International Workshop on Frontiers in Handwriting Recognition, PP.

[35] Y. Le Cun, O. Mattan, B. Boser, J.S. Denker et al, “Handwritten Zip Code Recognition with Multilayer Networks”,

Proceedings of International Conference on Pattern Recognition, Atlantic City, NJ, USA, Vol. 2, pp. 35

[36] V. Bansal, “Integrating Knowledge Sources

[37] B. B. Chaudhuri and U. Pal, “A Complete Printed Bangla OCR System”, Pattern Recognition, Vol. 31, No. 5, pp. 531

549,1998.

[38] B. B. Chaudhuri and U. Pal, “An OCR System to Read Tw

Proceedings of the Fourth International Conference on Document Analysis and Recognition, pp. 1011

[39] G. S. Lehal and C. Singh, “Feature Extraction and Classification for OCR of Gurmuk

12,1999.

[40] B. B. Chaudhuri, U. Pal and M. Mitra, “Automatic Recognition of Printed Oriya Script”, Sadhana, Vol. 27, pp. 23

[41] T. V. Ashwin and P S Sastry, “A Font and Size

Vector Machines”, Sadhana, Vol. 27, Part 1, pp. 35

[42] S. Sural and P.K. Das, “An MLP using Hough Transform Based Fuzzy Feature Extraction for Bengali Script Recognition”,

Pattern Recognition Letter, Vol. 20, pp. 771

[43] U. Garain and B. B. Chaudhuri, “Segmentation of Touching Characters in Printed Devanagari and Bangla Scripts using

Fuzzy Multifactorial Analysis”, IEEE Transactions on Systems, Man, Cybernatics

32, No. 4,2002.

[44] V. Bansal and R. M. K. Sinha, “Segmentation of Touching and Fused Devanagari Characters”, Pattern Recognition, Vol.

32, pp. 875-893 (2002).

[45] A. Bishnu and B. B. Chaudhuri, “Segmentation of Bangla Handwritten Text into

Following”, Proceedings of Fifth International Conference on Document Analysis and Recognition, Bangalore, India, pp.

402-405,1999.

[46] S. Kumar, “Theoretical Analysis of Devanagari Handprinted Characters

Journal of Computational Intelligence Research, Volume 6, Number 2, pp. 185

[47] R. Bajaj , L. Dey and S. Chaudhury, “Devanagari numeral recognition by combining decision of multiple connectionist

classifiers”, Sadhna, vol. 27, Part 1, pp. 59

[48] U. Pal and B. B. Chaudhuri, “Printed Devanagari script OCR system”, Vivek, vol. 10, pp. 12

[49] R. M. K. Sinha and H.N. Mahabala "Machine recognition of Devanagari script", IEEE transactions on syst

cybernetics, vol. SHC-9, no.8, pp. 435-

[50] K. Jayanthit, A. Suzukit , H. Kanait , Y. Kawasoej, M. Kimurat and K. Kido, “Devanagari character recognition using

structure analysis”, Fourth IEEE region 10

Satish Kumar




ato, M. Suzuki, S. Omachi, H. Aso and Y. Nemoto, “A Handwritten Character Recognition System using Directional


S. Britto Jr., R. Sabourin, F. Bortolozzi and C. Y. Suen, “Foreground and Background Information in an HMM


ognition, Kokubunji, Tokyo, Japan, pp. 371-376(2004).

T. W. Bhowmik, U. Bhattacharya and S.K. Parui, “Recognition of Bangla Handwritten Characters using an MLP Classifier

Based on Stroke Features”, Proceedings of ICONIP, Kolkata, India, pp. 814-819(2004).

K. Roy, S. Vajda, U. Pal and B.B. Chaudhuri, “ A System Towards Indian Postal Automation”, Proceedings of the Ninth

International Workshop on Frontiers in Handwriting Recognition, PP. 580 - 585 , 2004.

Boser, J.S. Denker et al, “Handwritten Zip Code Recognition with Multilayer Networks”,

Proceedings of International Conference on Pattern Recognition, Atlantic City, NJ, USA, Vol. 2, pp. 35

V. Bansal, “Integrating Knowledge Sources in Devanagari Text Recognition System”, Ph. D Thesis, 1996.

B. B. Chaudhuri and U. Pal, “A Complete Printed Bangla OCR System”, Pattern Recognition, Vol. 31, No. 5, pp. 531

Chaudhuri and U. Pal, “An OCR System to Read Two Indian Language Scripts: Bangla and Devanagari (Hindi)”,

of the Fourth International Conference on Document Analysis and Recognition, pp. 1011

G. S. Lehal and C. Singh, “Feature Extraction and Classification for OCR of Gurmukhi Script”, Vivek, Vol. 12, pp. 2

B. B. Chaudhuri, U. Pal and M. Mitra, “Automatic Recognition of Printed Oriya Script”, Sadhana, Vol. 27, pp. 23

T. V. Ashwin and P S Sastry, “A Font and Size-independent OCR System for Printed Kannada Documents using Support

Vector Machines”, Sadhana, Vol. 27, Part 1, pp. 35–58,2002.

S. Sural and P.K. Das, “An MLP using Hough Transform Based Fuzzy Feature Extraction for Bengali Script Recognition”,

l. 20, pp. 771–782,1999.

“Segmentation of Touching Characters in Printed Devanagari and Bangla Scripts using

Fuzzy Multifactorial Analysis”, IEEE Transactions on Systems, Man, Cybernatics- Part C: Applications and Revi

V. Bansal and R. M. K. Sinha, “Segmentation of Touching and Fused Devanagari Characters”, Pattern Recognition, Vol.

A. Bishnu and B. B. Chaudhuri, “Segmentation of Bangla Handwritten Text into Characters by Recursive Contour


ar, “Theoretical Analysis of Devanagari Handprinted Characters—A Recognition Perspective”,

Intelligence Research, Volume 6, Number 2, pp. 185–193, 2010.

R. Bajaj , L. Dey and S. Chaudhury, “Devanagari numeral recognition by combining decision of multiple connectionist

dhna, vol. 27, Part 1, pp. 59-72, 2002.

U. Pal and B. B. Chaudhuri, “Printed Devanagari script OCR system”, Vivek, vol. 10, pp. 12-24, 1997.

R. M. K. Sinha and H.N. Mahabala "Machine recognition of Devanagari script", IEEE transactions on syst

441, 1979.

K. Jayanthit, A. Suzukit , H. Kanait , Y. Kawasoej, M. Kimurat and K. Kido, “Devanagari character recognition using

structure analysis”, Fourth IEEE region 10th

international conference, Bombay, India, pp. 363-366, 1989.

190

ato, M. Suzuki, S. Omachi, H. Aso and Y. Nemoto, “A Handwritten Character Recognition System using Directional


S. Britto Jr., R. Sabourin, F. Bortolozzi and C. Y. Suen, “Foreground and Background Information in an HMM-based


T. W. Bhowmik, U. Bhattacharya and S.K. Parui, “Recognition of Bangla Handwritten Characters using an MLP Classifier

K. Roy, S. Vajda, U. Pal and B.B. Chaudhuri, “ A System Towards Indian Postal Automation”, Proceedings of the Ninth

Boser, J.S. Denker et al, “Handwritten Zip Code Recognition with Multilayer Networks”,

Proceedings of International Conference on Pattern Recognition, Atlantic City, NJ, USA, Vol. 2, pp. 35-40(1990).

in Devanagari Text Recognition System”, Ph. D Thesis, 1996.

B. B. Chaudhuri and U. Pal, “A Complete Printed Bangla OCR System”, Pattern Recognition, Vol. 31, No. 5, pp. 531-

Scripts: Bangla and Devanagari (Hindi)”,

of the Fourth International Conference on Document Analysis and Recognition, pp. 1011–1016,1997.

hi Script”, Vivek, Vol. 12, pp. 2–

B. B. Chaudhuri, U. Pal and M. Mitra, “Automatic Recognition of Printed Oriya Script”, Sadhana, Vol. 27, pp. 23-34,2002.

ted Kannada Documents using Support

S. Sural and P.K. Das, “An MLP using Hough Transform Based Fuzzy Feature Extraction for Bengali Script Recognition”,

“Segmentation of Touching Characters in Printed Devanagari and Bangla Scripts using

Part C: Applications and Reviews, Vol.

V. Bansal and R. M. K. Sinha, “Segmentation of Touching and Fused Devanagari Characters”, Pattern Recognition, Vol.

Characters by Recursive Contour


Perspective”, International

R. Bajaj , L. Dey and S. Chaudhury, “Devanagari numeral recognition by combining decision of multiple connectionist

24, 1997.

R. M. K. Sinha and H.N. Mahabala "Machine recognition of Devanagari script", IEEE transactions on systems, man, and

K. Jayanthit, A. Suzukit , H. Kanait , Y. Kawasoej, M. Kimurat and K. Kido, “Devanagari character recognition using

366, 1989.




[51] I. K. Sethi and B. Chatterjee, Machine recognition of constrained hand

2, pp. 69-75, 1977.

[52] S. D. Connel, R.M.K. Sinha and A. K. Jain, “Recognition of unconstrained on

the international conference on pattern recognition,

[53] U. Pal, N. Sharma, T. Wakabayashi, F. Kimura, "Off

Proceedings of 9th

international conference on document analysis and recognition , vol. 1, pp. 496

[54] N. Sharma, U. Pal et al ,”Recognition of off

2006, LNCS 4338, pp. 805-816, 2006.

[55] P. S. Deshpande, L. Malik and S. Arora, “Fine classification & recogntiton of handwritten Devanagari characters with

regular expression & minimum edit distance method”, Journal of computers, vol. 3, no. 5, pp. 11

[56] S. Kumar, “Devanagari Hand-printed Character Recognition using Multiple Features and Multi

International Journal of Computer Information Systems and

2010.

[57] S. Kumar, “Performance comparison of features on Devanagari hand

in engineering, vol. 1, no. 2, pp. 33-37, 2009.

[58] B. Shaw, S. K. Parui and M. Shridhar, “Off

approach”, IEEE , 2008.

[59] S. K. Parui and B. Shaw, “Off-line Devanagari handwritten word recognition: An HMM based approach”, Proceedings of

PReMI-2007(Springer), LNCS-4815, pp. 528

[60] B. Singh , A. Mittal, M.A. Ansari and D. Ghosh

Word Recognition: A Curvelet Transform Based Approach, International Journal of Computer S

Vol. 3, No. 4,pp. 1658-1665, 2011.

[61] A. S R amteke, andM. E Rane, Offline Handwritten Devanagari Script Segmentation, International Journal Of Scientific &

Technology Research Volume 1, Issue 4, MAY 2012

[62] R. Jayadevan , S. R. Kolhe , P. M. Patil and Umapada Pal,

Devanagari Legal Amount Words, International Conference on Document Analysis and Recognition, pp. 304

[63] S.G. Oval and S. Shirawale, Recognizing Devanagri Words using Recurrent Neural Network, Proceedings of the 3

International Conference on Frontiers of Intelligence Computing: Theory and Applications, pp. 413

[64] D. Ghosh, T. Dube and A. P. Shivaprasad, Script Recogni

Machine Intelligence, Vol 32, No. 12, pp 2142

[65] U. Pal, B.B. Chaudhuri, “Indian Script Character Recognition: A Survey”, Pattern Recognition, Vol. 37, pp. 1887

2004.

[66] R. C. Gonzalez and R. E. Woods, “Digital Image Processing”, 2nd Ed., Pearson Education.

[67] S. Kumar, “An Analysis of Irregularities in Devanagari Script

International Journal on Computer Science and Engin

Satish Kumar




I. K. Sethi and B. Chatterjee, Machine recognition of constrained hand-printed Devanagari, Pattern Recognition, vol. 9, no.

S. D. Connel, R.M.K. Sinha and A. K. Jain, “Recognition of unconstrained on-Line Devanagari characters”, Proceedings of

the international conference on pattern recognition, Barcelona, Spain, vol. 2, pp. 368-371, 2000.

i, F. Kimura, "Off-line handwritten character recognition of Devanagari script",

international conference on document analysis and recognition , vol. 1, pp. 496

N. Sharma, U. Pal et al ,”Recognition of off-line handwritten Devanagari characters using quadratic classifier”,

P. S. Deshpande, L. Malik and S. Arora, “Fine classification & recogntiton of handwritten Devanagari characters with

distance method”, Journal of computers, vol. 3, no. 5, pp. 11

printed Character Recognition using Multiple Features and Multi

International Journal of Computer Information Systems and Industrial Management Applications

Kumar, “Performance comparison of features on Devanagari hand-printed dataset”, International journal of recent trends

37, 2009.

K. Parui and M. Shridhar, “Off-line handwritten Devanagari word recognition: A segmentation based

line Devanagari handwritten word recognition: An HMM based approach”, Proceedings of

4815, pp. 528-535, 2007.

A. Mittal, M.A. Ansari and D. Ghosh, Handwritten Devanagari Word Recognition: A, Handwritten Devanagari

Word Recognition: A Curvelet Transform Based Approach, International Journal of Computer S

Offline Handwritten Devanagari Script Segmentation, International Journal Of Scientific &

Technology Research Volume 1, Issue 4, MAY 2012

R. Jayadevan , S. R. Kolhe , P. M. Patil and Umapada Pal, Database Development and Recognition of Handwritten

International Conference on Document Analysis and Recognition, pp. 304

, Recognizing Devanagri Words using Recurrent Neural Network, Proceedings of the 3

International Conference on Frontiers of Intelligence Computing: Theory and Applications, pp. 413

D. Ghosh, T. Dube and A. P. Shivaprasad, Script Recognition – A Review”, IEEE Transaction on Pattern Analysis and

Machine Intelligence, Vol 32, No. 12, pp 2142-2161, 2010.

U. Pal, B.B. Chaudhuri, “Indian Script Character Recognition: A Survey”, Pattern Recognition, Vol. 37, pp. 1887

R. C. Gonzalez and R. E. Woods, “Digital Image Processing”, 2nd Ed., Pearson Education.

An Analysis of Irregularities in Devanagari Script Writing – A Machine Recognition Perspective

l on Computer Science and Engineering Vol. 2, No. 2, pp. 274-279, 2010.

191

printed Devanagari, Pattern Recognition, vol. 9, no.

Line Devanagari characters”, Proceedings of

line handwritten character recognition of Devanagari script",

international conference on document analysis and recognition , vol. 1, pp. 496-500, 2007.

ten Devanagari characters using quadratic classifier”, ICVGIP

P. S. Deshpande, L. Malik and S. Arora, “Fine classification & recogntiton of handwritten Devanagari characters with

distance method”, Journal of computers, vol. 3, no. 5, pp. 11-17, May 2008.

printed Character Recognition using Multiple Features and Multi-stage Classifier”,

nt Applications , Vol.2, pp.039-055,

printed dataset”, International journal of recent trends

line handwritten Devanagari word recognition: A segmentation based

line Devanagari handwritten word recognition: An HMM based approach”, Proceedings of

, Handwritten Devanagari Word Recognition: A, Handwritten Devanagari

Word Recognition: A Curvelet Transform Based Approach, International Journal of Computer Science and Engineering,

Offline Handwritten Devanagari Script Segmentation, International Journal Of Scientific &

Database Development and Recognition of Handwritten

International Conference on Document Analysis and Recognition, pp. 304-308, 2011.

, Recognizing Devanagri Words using Recurrent Neural Network, Proceedings of the 3rd

International Conference on Frontiers of Intelligence Computing: Theory and Applications, pp. 413-421, 2014.

A Review”, IEEE Transaction on Pattern Analysis and

U. Pal, B.B. Chaudhuri, “Indian Script Character Recognition: A Survey”, Pattern Recognition, Vol. 37, pp. 1887-1899,

A Machine Recognition Perspective”,

Machine Recognition of Devanagari Hand -printed Script...

Documents

Transcript of Machine Recognition of Devanagari Hand -printed Script...