Jump to content

Pattern matching

From Simple English Wikipedia, the free encyclopedia
Revision as of 21:34, 24 May 2008 by Ur~simplewiki (talk | changes)

Pattern matching is family of tools for finding similar objects in diferent sources.
In image processing, the pattern matching is used for locate small image (called model) in a bigger one (called target image). The simple way is to move model in traget image and measure similarity in each position. The position with higher value of similarity will be accepted as result. The normolized correlation is used as common metric for similarity. The common way for reduce amount of calculation is using image pyramid. When using image pyramid, the serach is started on redused target image, and results are imporved in the next, more detailed one. This process is similar to processes in human vision.