Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

29
Noura A. Semary 1 , Sondos M. Fadl, Magda S. Essa, Ahmed F. Gad Faculty of Computers and Information, Menoufia University, Shebeen El-Kom, Menoufia, Egypt Member in Scientific Research Group in Egypt (SRGE) [email protected] 1 IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Transcript of Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Page 1: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Noura A. Semary1, Sondos M. Fadl, Magda S. Essa, Ahmed F. GadFaculty of Computers and Information, Menoufia University,

Shebeen El-Kom, Menoufia, EgyptMember in Scientific Research Group in Egypt (SRGE)

[email protected]

1IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 2: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Introduction

Related Works

Proposed Currency Recognition System

Experiments Results

Conclusion

2IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 3: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

The World Health Organization estimated the number of visually impaired at the global level, based on the latest studies, that there are about 285 million people who suffer from visual impairment worldwide, of whom 246 million people have visual impairment and 39 million people are totally blind

3IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 4: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

One of the main problems faced by people with visual disabilities is the inability to recognize the paper currencies due to the similarity of paper texture and size between the different categories.

4IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 5: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Types of Intelligent currency recognition systems

Scanner-based

Scan whole the paper

Specific position and

direction

Money counters

Camera-based

Capture any part of the

paper

Any direction

Mobile devices

5IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 6: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

6IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 7: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

7IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 8: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

8IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 9: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Objectives

Capturing any part of the money paper in any direction.

Simplify the technique of feature extraction and matching (for run time application)

9IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 10: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Block diagram

Image acquisition

Pre-processing

Segmentation

Histogram equalization

Template

Matching

ROI extraction

10IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 11: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Image Acquisition and Preprocessing

The image is captured by digital camera as RGB image and is converted to gray scale version by intensity equation 1.

Image acquisition

Pre-processing

Segmentation

Histogram equalization

Template

Matching

ROI extraction

𝐼 = (𝑅 + 𝐺 + 𝐵)/3 (1)

11IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 12: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Image Acquisition and Preprocessing

Gaussian blurring is used for noise removal.

Image acquisition

Pre-processing

Segmentation

Histogram equalization

Template

Matching

ROI extraction

𝐺 𝑥, 𝑦

=1

2𝜋𝜎2𝑒−𝑥2+𝑦2

2𝜎2 (2)12IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 13: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Segmentation

Convert the image into binary image by Otsu thresholding

Image acquisition

Pre-processing

Segmentation

Histogram equalization

Template

Matching

ROI extraction

𝐺 𝑥, 𝑦 = 1 𝑖𝑓 𝐺(𝑥, 𝑦) ≥ 𝑡ℎ0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

(3)

13IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 14: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Holes Filling

Morphological holes filling is used for extract the whole paper part

Image acquisition

Pre-processing

Segmentation

Histogram equalization

Template

Matching

ROI extraction

14IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 15: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Cropping and Histogram equalization

Image acquisition

Pre-processing

Segmentation

Histogram equalization

Template

Matching

ROI extraction

15IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 16: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Region of Interest (ROI) extraction The extracted mony paper is

resized to the dataset height. if the currency paper size is M × N

,then the extracted paper region will be resized by M × N/M.

Then a block of size r × c is selected for time acceleration

Image acquisition

Pre-processing

Segmentation

Histogram equalization

Template

Matching

ROI extraction

16IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 17: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Template Matching

correlation is performed to measure the similarity between images in the database and the ROI part

Image acquisition

Pre-processing

Segmentation

Histogram equalization

Template

Matching

ROI extraction

𝑐𝑜(𝑢, 𝑣) = 𝑥,𝑦 𝑓 𝑥,𝑦 −𝑓′𝑢,𝑝 𝑡 𝑥−𝑢,𝑦−𝑢 −𝑡′

𝑥,𝑦 𝑓 𝑥,𝑦 −𝑓′𝑢,𝑝2 𝑥,𝑦 𝑡 𝑥−𝑢,𝑦−𝑢 −𝑡′ 2

0.5 (4)

17IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 18: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Matlab Android

18IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 19: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Egyptian paper money of values 1, 5, 10, 20, 50, 100 and 200 have been scanned for both sides and saved in a dataset.

19IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 20: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

120 test images have been examined, 20 for each class.

20IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 21: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

For Matlab version, all the images has been captured offline with a limitation of having both the horizontal edges of the paper in the image.

For the Android system an automatic snapshot function is used to detect the upper and lower horizontal edges before automatic capturing. Hough line detection transform is utilized to obtain the horizontal edges.

21IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 22: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Once both horizontal edges are detected a vibration effect is occurred to notify the user of correct money detection.

After a while, a sound pronounces the correct value in Arabic is hear.

22IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 23: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

the application running on Samsun GT-N800 tablet testing 5-pounds mony paper.

Click to hear the voice of recognizes paper

23IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 24: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

System Evaluation

The system has been evaluated under the Matlabsystem using accuracy measure.

𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 =𝑇𝑟𝑢𝑒

𝑇𝑟𝑢𝑒+𝐹𝑎𝑙𝑠𝑒× 100 (5)

24IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 25: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

5 pound 10 pound 20 pound 50 pound 100 pound

200 pound

Quantity 20 20 20 20 20 20True 18 17 20 16 19 17False 2 3 0 4 1 3

Accuracy (%) 90 85 100 80 95 85

The average accuracy was 89%. 25IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 26: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Simple Egyptian currency recognition system based on template matching has been proposed .

Simple image processing techniques like thresholding, noise removal, histogram equalization and segmentation are used to extract the ROI and facilitate the template matching procedure.

Correlation is used to find the ROI in the dataset images.

The system has been written in Matlab and OpenCV library under Android platform as well.

26IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 27: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

The Matlab system considers offline captured images while the Android considers automatic horizontal edges detection for best frame capturing.

The system average accuracy for 120 test image was 89%.

27IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 28: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

Enhancing the accuracy by building a model of features for each currency class.

Working now on using local features instead of template matching

Enhancing the best frame to be processed for runtime application

Adding more currencies to the database.

28IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015

Page 29: Currency Recognition System for Visually Impaired: Egyptian Banknote as a Study Case (ICTA2015)

For farther contact

Email: [email protected]

FB: https://www.facebook.com/Dr.NouraSemary

29IEEE 2015 International Conference on Information and Communication technology and Accessibility, ICTA 2015