Lecture_04_Image_Search
-
Upload
machinelearner -
Category
Documents
-
view
220 -
download
0
Transcript of Lecture_04_Image_Search
-
8/8/2019 Lecture_04_Image_Search
1/27
Image Searches,
Abstraction, Invariance36-350: Data Mining
2 September 2009
1
-
8/8/2019 Lecture_04_Image_Search
2/27
Medical: x-rays, brain imaging, histology (dothese look like cancerous cells?)
Satellite imagery
Fingerprints
Finding illustrations for lectures...
2
-
8/8/2019 Lecture_04_Image_Search
3/27
Searching for Images by
Searching for Text Assume theres text accompanying the images
(annotation)
tags
Search those text records with the query phrase
Take images which appear close to the queryphrase on highly-ranked records
This how Google does it
3
-
8/8/2019 Lecture_04_Image_Search
4/27
Sometimesthis worksperfectly...
4
-
8/8/2019 Lecture_04_Image_Search
5/27
...and
sometimesit doesnt;depends on
the text!
5
-
8/8/2019 Lecture_04_Image_Search
6/27
Searching for images by
representing images
For text, we only cared about features, and
only worked with feature vectors
Define numerical features for images andeverything carries over
Abstraction
6
-
8/8/2019 Lecture_04_Image_Search
7/27
Abstraction
Remove some of the details but keep others
Kept details = features Then act on abstracta
Hopes:
Simplifies problem Lets you treat many problems similarly
7
-
8/8/2019 Lecture_04_Image_Search
8/27
Abstract level: feature vectors
Concrete level: meaningful objects
Text 1 Text 2 Text 3
BoW
BoW
v1 v2 v3 v4 v5 v6
Similaritymatching
DimensionalityReduction
Classification Clustering
BoW
BoW
Text 4 Text 5 Text 6
BoW
BoW
etc.
8
-
8/8/2019 Lecture_04_Image_Search
9/27
Concrete level: meaningful objects
Text 1 Text 2 Text 3
Topics
v1 v2 v3 v4 v5 v6
Topics
Text 4 Text 5 Text 6
Topics
Topics
Topics
Topics
Abstract level: feature vectors
Similaritymatching
DimensionalityReduction
Classification Clustering etc.
9
-
8/8/2019 Lecture_04_Image_Search
10/27
Concrete level: meaningful objects
Bitmap
v1 v2 v3 v4 v5 v6
Pic. 1 Pic. 2 Pic. 3 Pic. 4 Pic. 5 Pic.6
Bitmap
Bitmap
Bitmap
Bitmap
Bitmap
Abstract level: feature vectors
Similaritymatching
DimensionalityReduction
Classification Clustering etc.
10
-
8/8/2019 Lecture_04_Image_Search
11/27
Concrete level: meaningful objects
v1 v2 v3 v4 v5 v6
Pic. 1 Pic. 2 Pic. 3 Pic. 4 Pic. 5 Pic.6
Bagofcolors
Bagofcolors
Bagofcolors
Bagofcolors
Bagofcolors
Bagofcolors
Abstract level: feature vectors
Similaritymatching
DimensionalityReduction
Classification Clustering etc.
11
-
8/8/2019 Lecture_04_Image_Search
12/27
Concrete level: meaningful objects
Network 6
Motifs
v1 v2 v3 v4 v5 v6
Network 1 Network 2 Network 3 Network 4 Network 5
Motifs
Motifs
Motifs
Motifs
Motifs
Abstract level: feature vectors
Similaritymatching
DimensionalityReduction
Classification Clustering etc.
12
-
8/8/2019 Lecture_04_Image_Search
13/27
-
8/8/2019 Lecture_04_Image_Search
14/27
Concrete level: meaningful objects
Text 1 Text 2 Text 3
BoW
BoW
Topics
Pic. 1 Pic. 2
Social
Network
v1 v2 v3 v4 v5 v6
Bagofcolors
Bitmap
Motifs
Abstract level: feature vectors
Similaritymatching
DimensionalityReduction
Classification Clustering etc.
14
-
8/8/2019 Lecture_04_Image_Search
15/27
flower1 flower2 flower3
tiger1 tiger2 tiger3
ocean1 ocean2 ocean315
-
8/8/2019 Lecture_04_Image_Search
16/27
Euclidean Distance of
Images Image is MxN pixels, each with 3 color
components, so a 3MN vector
Euclidean distance possible, and OK forsome kinds of noise-removal
but hopeless even at grouping flower1 with
flower2
or slight changes in perspective, lighting...
16
-
8/8/2019 Lecture_04_Image_Search
17/27
Bag of Colors
If it works, try it some more
For each possible color, count how manypixels there are of that color
Use Euclidean distance on color-countvectors
Too many colors, so quantize them down toa manageable number (like stemming, orcombining synonyms)
17
-
8/8/2019 Lecture_04_Image_Search
18/27
flower1
flower2
flower3
flower4
flower5
flower6
flower7
flower8
flower9
tiger1
tiger2
tiger3
tiger4
tiger5
tiger6
tiger7
tiger8
tiger9
ocean1
ocean2
ocean3
ocean4
ocean5
ocean6
ocean7
ocean7
ocean6
ocean5
ocean4
ocean3
ocean2
ocean1
tiger9
tiger8
tiger7
tiger6tiger5
tiger4
tiger3
tiger2
tiger1
flower9
flower8
flower7
flower6
flower5
flower4
flower3
flower2
flower1
1.0 0.5 0.0 0.5 1.0
1.0
0.5
0.0
0.5
1.0
V1
V2
flower1
flower2
flower3
flower4
flower5
flower6
flower7
flower8
flower9
tiger1tiger2
tiger3
tiger4
tiger5
tiger6
tiger7tiger8
tiger9
ocean1
ocean2
ocean3ocean4
ocean5 ocean6
ocean7
flower ocean tiger
Multidimensional scaling
Distances between images MDS plot of images18
-
8/8/2019 Lecture_04_Image_Search
19/27
Representation and
Invariance
Invariances of a representation = how can wechange the underlying object without changingthe representation?
What differences does the representation ignore?
19
-
8/8/2019 Lecture_04_Image_Search
20/27
Invariants of bags of words
Punctuation and word order
Universal words (exact count of the, of,to, ...), if using inverse document frequency
Word-endings, if using stemming
Grammar, context, word proximity ... Send lawyers, guns and money vs.Sending the Guns lawyers for themoney
20
-
8/8/2019 Lecture_04_Image_Search
21/27
-
8/8/2019 Lecture_04_Image_Search
22/27
Same color counts, different textures
22
-
8/8/2019 Lecture_04_Image_Search
23/27
Non-invariants
Lighting, shadows
Occlusion, 3D effects Blurring
There are good ways to deal with blur(from astronomy)
but full vision is very, very hard
23
-
8/8/2019 Lecture_04_Image_Search
24/27
Breaking an invariance is easy
e.g., add features for textures
or sub-divide the image and do color-counts on each part
Adding invariances is hard
often need to go back to scratch andchose a different representation
24
-
8/8/2019 Lecture_04_Image_Search
25/27
Similaritysearch
with real
imagesfrom the
web(retrievr,
see notes)
25
-
8/8/2019 Lecture_04_Image_Search
26/27
26
-
8/8/2019 Lecture_04_Image_Search
27/27
Typically works better with more restricteddomains (actually pretty good for medicalimages)