Image Retrieval Based on Color and Texture Features

Currently, content-based color image retrieval technology is widely used, however, if we consider using single visual feature only to characterize image, the effect is not perfect. This paper is to overcome the above drawbacks, and have proposed using multi-feature approach to represent a color image, and then adding a slight noise on the different categories of images to match, finally, the results of the multi-feature matching defined as J4.8 decision trees properties' input to train correspondingly. Experimental results have shown that color image retrieval accuracy of this method is higher, the operation is small, and it will be robust for the slight noise. Therefore, it is suitable for different types of color image retrieval.