cs4495 6495
play

CS4495/6495 Introduction to Computer Vision 4A-L2 Finding corners - PowerPoint PPT Presentation

CS4495/6495 Introduction to Computer Vision 4A-L2 Finding corners Feature points Characteristics of good features Repeatability/Precision The same feature can be found in several images despite geometric and photometric transformations


  1. CS4495/6495 Introduction to Computer Vision 4A-L2 Finding corners

  2. Feature points

  3. Characteristics of good features Repeatability/Precision • The same feature can be found in several images despite geometric and photometric transformations

  4. Characteristics of good features Saliency/Matchability • Each feature has a distinctive description

  5. Characteristics of good features Compactness and efficiency • Many fewer features than image pixels

  6. Characteristics of good features Locality • A feature occupies a relatively small area of the image; robust to clutter and occlusion

  7. Corner Detection: Basic Idea “corner”: “edge”: “flat” region: significant change no change no change in in all directions along the edge all directions with small shift direction Source: A. Efros

  8. Finding Corners • Key property: in the region around a corner, image gradient has two or more dominant directions

  9. Finding Corners C. Harris and M. Stephens. "A Combined Corner and Edge Detector,” Proceedings of the 4th Alvey Vision Conference : 1988

  10. Corner Detection: Mathematics Change in appearance for the shift [ u,v ]:    2     E u v w x y I x u y v I x y ( , ) ( , ) ( , ) ( , ) x y , Window Shifted Intensity function intensity Window function w(x,y) = or 1 in window, Gaussian 0 outside Source: R. Szeliski

  11. Corner Detection: Mathematics Change in appearance for the shift [ u,v ]:    2     E u v w x y I x u y v I x y ( , ) ( , ) ( , ) ( , ) x y , I ( x , y ) E ( u , v ) E (3,2) E (0,0)

  12. Corner Detection: Mathematics Change in appearance for the shift [ u,v ]:    2     E u v w x y I x u y v I x y ( , ) ( , ) ( , ) ( , ) x y , We want to find out how this function behaves for small shifts (u,v near 0,0)

  13. Corner Detection: Mathematics Change in appearance for the shift [ u,v ]:    2     E u v w x y I x u y v I x y ( , ) ( , ) ( , ) ( , ) x y , Second-order Taylor expansion of E ( u , v ) about (0,0) (local quadratic approximation for small u,v ):

  14. Corner Detection: Mathematics Change in appearance for the shift [ u,v ]:    2     E u v w x y I x u y v I x y ( , ) ( , ) ( , ) ( , ) x y , 2 d F d F (0 ) 1 (0 )       2 F x F x x ( ) (0 ) · · 2 d x d x 2

  15. Corner Detection: Mathematics Change in appearance for the shift [ u,v ]:    2     E u v w x y I x u y v I x y ( , ) ( , ) ( , ) ( , ) x y , 2 d F d F (0 ) 1 (0 )       2 F x F x x ( ) (0 ) · · 2 d x d x 2       E E E u (0, 0 ) (0, 0 ) (0, 0 ) 1 u u u u v    E u v E u v u v ( , ) (0, 0 ) [ ] [ ]       E E E v (0, 0 ) (0, 0 ) (0, 0 )     2   v u v vv

  16.    2     E u v w x y I x u y v I x y ( , ) ( , ) ( , ) ( , ) x y , Second-order Taylor expansion of E ( u , v ) about (0,0):      1  E E E (0, 0 ) (0, 0 ) (0, 0 ) u    E u v E u v u v ( , ) (0, 0 ) [ ] [ ] u u u u v       E E E v (0, 0 ) (0, 0 ) (0, 0 )       2 v u v vv

  17.    2     E u v w x y I x u y v I x y ( , ) ( , ) ( , ) ( , ) x y , Second-order Taylor expansion of E ( u , v ) about (0,0):      1  E E E (0, 0 ) (0, 0 ) (0, 0 ) u    E u v E u v u v ( , ) (0, 0 ) [ ] [ ] u u u u v       E E E v (0, 0 ) (0, 0 ) (0, 0 )       2 v u v vv Need these derivatives…

  18.    2     E u v w x y I x u y v I x y ( , ) ( , ) ( , ) ( , ) x y , Second-order Taylor expansion of E ( u , v ) about (0,0):          I E u u v w x y I x u y v I x y x u y v ( , ) 2 ( , ) ( , ) ( , ) ( , ) x x , y

  19.    2     E u v w x y I x u y v I x y ( , ) ( , ) ( , ) ( , ) x y , Second-order Taylor expansion of E ( u , v ) about (0,0):       I I E u u u v w x y x u y v x u y v ( , ) 2 ( , ) ( , ) ( , ) x x x , y          I w x y I x u y v I x y x u y v 2 ( , ) ( , ) ( , ) ( , ) xx x , y

  20.    2     E u v w x y I x u y v I x y ( , ) ( , ) ( , ) ( , ) x y , Second-order Taylor expansion of E ( u , v ) about (0,0):       I I E u v w x y x u y v x u y v ( , ) 2 ( , ) ( , ) ( , ) u v y x x y ,          I w x y I x u y v I x y x u y v 2 ( , ) ( , ) ( , ) ( , ) xy x y ,

  21. Second-order Taylor expansion of E ( u , v ) about (0,0):       1 E E E u (0, 0 ) (0, 0 ) (0, 0 )    E u v E u v u v ( , ) (0, 0 ) [ ] u [ ] u u u v       E E E v (0, 0 ) (0, 0 ) (0, 0 )       2 v u v vv          E u v w x y I x u y v I x y I x u y v ( , ) 2 ( , ) ( , ) ( , ) ( , ) u x x , y       E u v w x y I x u y v I x u y v ( , ) 2 ( , ) ( , ) ( , ) u u x x x y ,          w x y I x u y v I x y I x u y v 2 ( , ) ( , ) ( , ) ( , ) xx x , y       E u v w x y I x u y v I x u y v ( , ) 2 ( , ) ( , ) ( , ) u v y x x y ,          w x y I x u y v I x y I x u y v 2 ( , ) ( , ) ( , ) ( , ) xy x y ,

  22. Evaluate E and its derivatives at (0,0) : = 0       1 E E E u (0, 0 ) (0, 0 ) (0, 0 )    E u v E u v u v ( , ) (0, 0 ) [ ] [ ] u u u u v       E E E v (0, 0 ) (0, 0 ) (0, 0 )       2 v u v vv      E w x y I x y I x y I x y (0, 0 ) 2 ( , ) ( , ) ( , ) ( , ) u x = 0 x , y   E w x y I x y I x y (0, 0 ) 2 ( , ) ( , ) ( , ) u u x x x y = 0 ,      w x y I x y I x y I x y 2 ( , ) ( , ) ( , ) ( , ) xx x , y   E w x y I x y I x y (0, 0 ) 2 ( , ) ( , ) ( , ) u v y x = 0 x y ,      w x y I x y I x y I x y 2 ( , ) ( , ) ( , ) ( , ) xy x y ,

  23. Second-order Taylor expansion of E ( u , v ) about (0,0):   1     E E E u (0, 0 ) (0, 0 ) (0, 0 )    E u v E u v u v ( , ) (0, 0 ) [ ] [ ] u u u u v       E E E v    ( 0, 0 )   (0, 0 ) (0, 0 )  2 v u v vv    E E w x y I x y I x y (0, 0 ) 0 (0, 0 ) 2 ( , ) ( , ) ( , ) u u x x x , y    E E w x y I x y I x y (0, 0 ) 0 (0, 0 ) 2 ( , ) ( , ) ( , ) vv y y u x , y    E E w x y I x y I x y (0, 0 ) 0 (0, 0 ) 2 ( , ) ( , ) ( , ) u v x y v x , y

  24. Second-order Taylor expansion of E ( u , v ) about (0,0):      2 w x y I x y w x y I x y I x y ( , ) ( , ) ( , ) ( , ) ( , )  u x x y  E u v u v  x y x y   ( , ) [ ] , ,    2 w x y I x y I x y w x y I x y v ( , ) ( , ) ( , ) ( , ) ( , )       x y y x y x y , ,    E E w x y I x y I x y (0, 0 ) 0 (0, 0 ) 2 ( , ) ( , ) ( , ) u u x x x , y    E E w x y I x y I x y (0, 0 ) 0 (0, 0 ) 2 ( , ) ( , ) ( , ) vv y y u x , y    E E w x y I x y I x y (0, 0 ) 0 (0, 0 ) 2 ( , ) ( , ) ( , ) u v x y v x , y

  25. Corner Detection: Mathematics The quadratic approximation simplifies to   u  E u v u v M ( , ) [ ]   v   where M is a second moment matrix computed from image derivatives:   2 I I I  x x y    M w x y ( , ) 2 I I I     x y , x y y

  26. The second moment matrix M:   2 I I I  x x y    M w x y ( , ) 2 I I I     x y , x y y Each product is a rank 1 2x2 Can be written (without the weight):         I I I I I   x   T       x x x y    M I I I I   ( )       x y I  I I I I        y x y y y

  27. Interpreting the second moment matrix The surface E ( u , v ) is locally approximated by a quadratic form.   u  E u v u v M ( , ) [ ]   v     2 I I I   M w x y x x y ( , )   2 I I I   x y , x y y

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend