fa13-cs188-lecture-22-1PP

0 dot product of two images vectors 1 usually

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: est neighbor for digits:   Take new image   Compare to all training images   Assign based on closest example 0 1   Encoding: image is vector of intensi)es: 2   What’s the similarity func)on? 0   Dot product of two images vectors? 1   Usually normalize vectors so ||x|| = 1   min = 0 (when?), max = 1 (when?) 2 Similarity Func)ons Basic Similarity   Many similari)es based on feature dot products:   If features are just the pixels:   Note: not all similari)es are of this form Invariant Metrics   BeOer similarity func)ons use knowledge about vision   Example: invariant metrics:   Similari)es are invariant under certain transforma)ons   Rota)on, scaling, transla)on, stroke ­thickness…   E.g:   16 x 16 = 256 pixels; a point in 256 ­dim space   These points have small similarity in R256 (why?)   How can we incorporate such invariances into our similari)es? This and next few slides adapted from Xiao Hu, UIUC Rota)on Invariant Metrics   Each example is now a curve in R256   Rota)on invariant similarity: s’=max s( r( ), r( ))   E.g. highest similarity between images’ rota)on lines Template Deforma)on   Deformable templates:         An “ideal” version of ea...
View Full Document

This note was uploaded on 12/22/2013 for the course CS 188 taught by Professor Staff during the Fall '08 term at University of California, Berkeley.

Ask a homework question - tutors are online