| Nowadays,with the unceasing progress of science and technology,the clothing e-commerce has developed rapidly,and the clothing images on the network have grown explosively.In order to achieve the goal of finding the required images more efficiently,this paper puts forward a method of clothing recognition based on multi-features fusion,which is based on the Content Based Image Retrieval.In this paper,we focus on the clothing image feature extraction technology.And the work is mainly as follows:First,the recent progress in the world of studies about the CBIR system is introduced,and some correlative algorithms are presented aiming at three low-level features of clothing images.As for the color feature extraction,we put forward an algorithm based on HSV and color moment.As for the texture feature extraction,we put forward an algorithm based on Fourier descriptor.Besides,on the shape feature extraction,we put forward an algorithm based on Hu moment invariants.Second,this paper studies the key technology of clothing image retrieval based on multi-features fusion.Because of the clothing image' containing high information,a single bottom visual feature can't express fully on the target image.Therefore,the effect of retrieval is very poor.This paper puts forward the method based on multi-features fusion to retrieval clothing image.And the experimental results show that our proposed algorithm has better retrieval performance than using a single bottom feature algorithm.Third,the author has designed and developed the clothing images retrieval system based on multi-features fusion.The platform of this system is Matlab2010a.And the system has seven function modules,such as select menu,setting path,inputting image to be retrieval,custom weights,the characters of the image to be retrieval and the search results.The performance of each function module is tested,and the validity of the method is demonstrated effectively by the clotthing images retrieval system. |
PDF Full Text Request