The problem of color space and histogram distance in histogram-based image retrieval is discussed, and the common frame of histogram -based image retrieval is proposed. Then eight kinds of image retrieval algorithms based on two color spaces and four histogram distance algorithms are created and also two objective metrics for evaluating their effectiveness are adopted. The results show that HSV color space is more effective than RGB color space in common. For histogram distance, EMD has showed better performance than other distance, but it has a relative greater computational burden. Comparing with EMD, histogram intersection distance needs less computational burden and only has slightly weak performance.
Infrared and Laser Engineering
Color image processing