1 / 25

Lecture 7

Lecture 7. Generalized Hough Transform Texture Segmentation. Generalized Hough Transform. Any fixed shape 2 parts: 1. Learn shape properties 2. Search for shape (target) in image. Learn Shape - “measurements”. 1) Choose an arbitrary “center” point

krikor
Download Presentation

Lecture 7

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Lecture 7 Generalized Hough Transform Texture Segmentation

  2. Generalized Hough Transform Any fixed shape 2 parts: 1. Learn shape properties 2. Search for shape (target) in image

  3. Learn Shape - “measurements” 1) Choose an arbitrary “center” point 2) At some edge points, draw line to center a - angle with x-axis r - distance to center q - gradient from edge detector

  4. q r1, a1 r2, a2 r3, a3 r4, a4 r5, a5 0...19 15, 180 15, 179 16, 177 13,176 14,175 20...39 17, 160 14, 159 18, 161 15, 162 14, 163 30…49 19, 165 20, 170 22, 167 18, 159 21, 161 … … … … … … 340...359 23, 105 24, 103 21,102 22, 104 20, 103 GHT: R-table From a relationship between q, (a,r)

  5. Search for Shapes For each edge element 1. Use q to find all (ai,ri) for that q 2. For each (ai,ri) xc = x + r cos (a) yc = y + r sin (a) 3. Look for Max (xc,yc) the object is there

  6. Search For Shape 180-a xc – x = r cos (a) and yc – y = r sin (a)

  7. Search for Shape 1. Array A[xcmin..xcmax, ycmin…ycmax] initialized to 0 2. For each edge point with gradient q(x,y) find each (a,r) from R-Table xc = x + r cos (a) yc = y + r sin (a) A[xc,yc]++; 3. Look for peaks in A each peak means object is there

  8. Case when target is scaled/rotated Scale factor S rotation factorb Create A of 4-dimension A[xc, yc, S, b] 0..359 b = 180 0.3,0.4,0.5,0.6,0.7,0.8,0.9,1,1.1,1.2,1.3,1.4,1.5 xc = x + r S cos (a+b) yc = y + r S sin (a+b) xc = x + r cos (a) yc = y + r sin (a)

  9. Case when target is scaled/rotated For each (a,r) from R-Table *For each S = .3 to 1.5 step .1 *For each b = 0 to 359 step 1 xc = x + r S cos (a+b) yc = y + r s sin (a+b) A[xc,yc,A,b]++; Look for peaks in 4-D Array Impractical!!

  10. Modified Genearlized Hough Transform Reduced 4D to 2D Allow Scale & Rotation

  11. Dq r1, a1, q1, L1 r2, a2, q2, L2 r3, a3, q3, L3 r4, a4, q4, L4 0...19 15,180,195,99 9,179,219,101 8,177,216,102 9,176,198,100 20...39 17,160,23,5 14,159,38,7 18,161,175,62 15,162,195,95 30…49 19,165,31,53 20,170,8,52 22,167,15,52 18,159,158,12 … … … … … 340..359 23,105,346,11 24,103,165,11 21,102,346,18 22,104,195,24 Modified Genearlized Hough Transform Gradient Direction is perpendicular to edge direction Dq (a, r, L, q)

  12. MGHT: Rotation/Scale Invariant Rotated Object Smaller Object Dqis constant bring out a, r, L, q LT - L of target Rotation factor = qT-q = b xc = x + r S cos (a+b) yc = y + r s sin (a+b) Scale factor = LT/L = S

  13. Image Querying Database Search for Circle shape Search for bulb shape

  14. Segmentation Group pixels that belong to same region with 4-connected color

  15. Texture Segmentation Segment images into texture regions

  16. Texture Segmentation Texture : - Statistical distribution of gray scale - Change in distribution change in texture Spatial Gray Level Distributed Matrix (SGLD) - Co-occurrence Matrix Given a direction, it gives probability distribution of gray scale in that direction

  17. Texture Segmentation (0,0) d = (1,1) Total=27 (0,1) (1,0) (1,1) P(0,0) = 6/27; P(0,1) = 0 P(1,0) = 6/27; P(1,1) = 15/27

  18. Texture Segmentation j x 16 i Gray-level co-occurrence matrix for d = (1,1) 5 x 5 image with three gray levels

  19. Texture Segmentation j j x (1/56) i i x (1/49) 8 x 8 checkerboard image Gray-level co-occurrence matrix for d = (1,0) Gray-level co-occurrence matrix for d = (1,1)

  20. Use of Texture 1) Segmentation 2) Search Split and Merge

  21. หา Co-occurrencematrixที่ d[3,3]และ d[-3,3] กรองเฉลี่ย LoG Food Inspection: Texture

More Related