features for handwriting recognition. | 2 the challenge “rappt jd 10 feb no 175, om machtiging om...
Post on 18-Dec-2015
219 views
TRANSCRIPT
| 3
Short processing pipeline
“machtiging”
Feature extraction
Classification
82,34,66,…0.12
“machtiging”
Learning
Preprocessing
› Goal: enhance the foreground while reducing other visual symptoms (stains, noise, pictures, ...)
› Methods:• Contrast stretching• Highpass filtering• Despeckling• Change color representation (RGB, HSV,
grayscale, black/white, …)• Remove selected connected components ()• …
| 6
| 9
Object of classification› Sentences› Words› Characters
(use grammar)(use dictionary)(use alphabet)
| 10
Object representations
› Image› Unordered vectors (in a coco)› Contour vectors› On-line vectors› Skeleton image› Skeleton vectors
(x, y)i
(x, y)k
(x, y)k
(x, y)k
I(x, y)
I(x, y)
| 11
A full processing pipeline
Segmentation
Normalization
Feature extraction
Classification
Preprocessing
| 12
Invariance
› Luminance / contrast› Position› Size› Rotation› Shear› Writer style› Ink thickness› …
| 13
Invariance by normalization
› Luminance / contrast› Position› Size› Rotation› Shear› Writer style› Ink thickness› …
Center on center of gravity
Contrast stretching
Scale to standard
size
| 14
Invariance by trying many deformations› Luminance / contrast› Position› Size› Rotation› Shear› Writer style› Ink thickness› …
Try different scale factors
Try different rotations
… and use the best recognition result
Try different deformation
s
| 15
Invariance by using invariant features
› Luminance / contrast› Position› Size› Rotation› Shear› Writer style› Ink thickness› …
Zernike invariant moments
| 16
A full processing pipeline
Segmentation
Normalization
Feature extraction
Classification
Preprocessing
82,34,66,…
| 21
Feature types
› Image itself› Statistical› Structural› Abstract
› Image (off-line) features (1—20)› Contour / on-line features (21 – 28)
| 34
"million" ==> convex:concave:3(north:concave) :(north:LOOP):concave:(north:LOOP) :concave:north :concave:HOLE :2(convex:concave)
(J.-C. Simon, 1989)
Feature 14: J.C. Simon (2/2)
| 40
Feature 17: Fourier transform (2/2)
Fig. 1 and 3 from: http://www.csse.uwa.edu.au/~wongt/matlab.html
Fig. 2 from: http://www.chemicool.com/definition/fourier_transform.html
| 41
Feature 18: Wavelet transform
From: http://www.regonaudio.com/Audio%20Measurement%20via%20Wavelets.html
| 42
Feature 19: Hu invariant moments
dxdyyxyxiM q
D
pqp ).,(,
0,0M area of the object
0,11,0 ,MM center of mass
Slide from: http://www.cedar.buffalo.edu/~govind/CSE717/lectures/CSE717_3.ppt
› Invariant for scale, position and rotation
› Derived from moments› Moments describe the image distribution with
respect to its axes › Works on (x, y) vectors
| 43
Feature 20: Zernike moments
From: Trier, O. D., Jain, A. K., and Taxt, T. (1996). Feature extraction methods for character recognition - a survey. Pattern Recognition,29:641–662.
| 44
Feature 21 – 28: Contour features
› (cos, sin) of running angle› (cos, sin) of running angular difference› Angular difference› Fourier transform› Ink density (horizontal or vertical)› Radon transform: (ink density, computed radially from
the c.o.g.)› Angular histogram› Curvature scale space ()
| 45
Feature 28: Curvature scale space
From: http://www.christine.oppe.info/blog/category/formen-und-farben/formenvergleich/
pos
itera
tion