论文部分内容阅读
Based on a class of complete orthogonal function system,V-system,this paper proposes a new kind of moment functions(called V-moment functions),and applies them to the shape retrieval.The V-moments are orthogonal,and involve only simple linear operations.The V-moments can be used to extract image features accurately,and the original image can be reconstructed with only a small amount of them.The V-moments have advantage in extracting features of image with complex boundaries since the V-system contains a great deal of discontinuous basis functions.Therefore,feature extraction of multi-boundary image using V-moments is very promising.This paper performs image retrieval based on their shape features.The results of retrieval experiment,conducted on benchmark database MPEG-7-shape-CE2,show that the algorithm proposed in this paper outperforms some classical moments including Zernike moments,Hu invariant moments,orthogonal Fourier-Mellin moments,Legendre moments and the geometric central moments in retrieval efficiency according to several evaluation indexes.