Abstract: Aiming at the current problem of single basic color features of image and insufficient information available, we proposed to transform the image into HSV color space. We first used the equalization histogram to enhance the spatial color pixels of the image after the spatial color conversion, and then used K-means clustering algorithm integrated with rgb2ind to brighten the balanced image, so as to enhance the extraction target and quantity of image color feature. The experimental results show that the proposed method is better than that of the ordinary spatial color image for feature extraction, and provides more retrieval information for color feature recognition.

Key words:  , HSV color space, mean clustering, feature extraction, recognition, retrieval