Color Image Retrieval Based on Multiple Features
Ning Ding · Jiangnan daxue xuebao. Ziran kexue ban · 2009
In the paper,a new method of content-based image retrieval is presented,which integrates color and shape features.Firstly,the color image is quantized and encoded based on the special disposal on the HSV color space and the normalized histograms are extracted from every pixel and its multi-neighbor color moments as color feature.Secondly,low frequency band of integer wavelet transform is segmented into blocks and encoded and the transition probability matrix is calculated as shape information.Finally,the similarities between color images are computed by using a combined index based on color feature and shape feature.The experimental results show that the proposed algorithm is more accurate and efficient in retrieving the user-interested images.