Proceedings PaperRotation Invariant Pattern Recognition
|Format||Member Price||Non-Member Price|
This paper reports on recent progress on invariant pattern recognition. Circular harmonic component (CHC) matched filters allow pattern recognition that is invariant under translations and rotations of the objects, and does not require segmentation of the object from its background. Now that the problem of the proper center for expansion of the filter has been solved, it has become easier to use such filters. The performance of the filter may be improved without using multiple filters by using composite CHC filters. When both target discrimination and sidelobe reduction are required, a composite filter that is an average between two composite filters must be used. The use of other invariant features is discussed.