I have developed an algorithm for classifying leaves based on local analysis of their boundary curves. The wavelet transform is used to generate a coefficient vector at each boundary point over several scales. The distributions of these vectors are compared to determine similarity between leaves. This method is meant to complement an already-implemented leaf classification system whose decisions rely entirely on global shape information, and the combined results of the local and global models are presented.

MidYear Progress Report and Presentation

Final Presentation , Final Report