Research of image retrieval based on color and shape features

摘要

A new image retrieval algorithm based on color of image blocks and moment invariant of image was presented. First of all, the image is converted to HSV color space, then the image is separated into sub-blocks, and it gets the color histogram in each block as the image color feature. Meanwhile, it uses seven HU moment invariant and eccentricity to describe shape feature. Finally, the similarity is calculated based on the color histogram of each block and moment invariant. Experiment results show this method is very effective.