Image retrieval using both color and local spatial feature histograms

A novel color image retrieval method using both color and local spatial feature histograms (CLSFH) is proposed. In CLSFH, the non-uniform quantized HSV color model is used; the mean and the standard deviation of the 5×5 neighbors of every pixel are calculated, and are used to generate the local mean histogram and the local standard deviation histogram; the directional difference unit of the 3×3 neighbors of every pixel is defined and computed, and is used to generate the local directional difference unit histogram. The three histograms and the color histogram are used as feature indexes to retrieve color images. Experimental results show that CLSFH has better performance than other color-spatial based methods for color images, especially for color images with relatively regular texture or structure characteristics.