CV in a Nutshell (I) Yi Li inNutshell.htm
Outline Introduction Image Formation Early vision Cognition in psychology Face recognition (I)
Paper discussion (1) Recognition by Components: – Segmentation: blocks, cylinders, wedges, cones – Curvature, collinearity, symmetry, parallel – Contour, surface, texture – Occlusion – Top-down or bottom-up? – Relation: pairwise?
Paper discussion (2) Face Recognition – extremely low-resolution images (Yang and Ma) – Occlusion (sparse) – Component vs hollistic – Context! – Facial expression (Cohn) – Illumination (Basri and Jacobs) – Specialized (Martinez)
Image Formation Geometry – image plane – projection Physics of light – Brightness / Illumination – Light source – Lambertian scattered such that the apparent brightness of the surface to an observer is the same regardless of the observer's angle of view (wiki) nikonweb.com
Lighting
Early Vision Early year of computer vision – David Marr (Psychologist, MIT) – Azriel Rosenfeld (Mathematician, Maryland) – Takeo Kanade (Engineer, CMU) Visual path – How early? – Recognition? – Saliency / gist scienceblogs.com
Illusion Aperture problem ment2.html ment2.html
Skeletonization
Template Matching Find transformation to align two images cost function Example: Chamfer Matching Example 2: Hough transform Example 3: Probabilistic voting
Deformable Shape matching – Shape Context Snake – Active Contour, Active Appearance Articulated objects – Scissor – Human body
Face Recognition Linear subspaces – PCA and LDA – Eigenfaces vs Fisherfaces
Q/A Introduction Image Formation Early vision Cognition in psychology Face recognition (I)