The Research Of Color-based Image Retrieval

With the development and popularity of digital and computer technology, more and more digital image resources are available and the need for image retrieval becomes more and more urgent, thus giving birth to content-based image retrieval technique. Color is one of the basic features of images, which is one of the most widely used visual features by image retrieval. Compared with other visual features, the extraction of color features is relatively simple. Color feature has also translation invariance, rotation invariance and scale invariance. It has become a powerful image analysis tool. This paper studied the existing image retrieval algorithms based on color features. On this basis, three improved algorithms are proposed.(l)The circular color histogram algorithm which put forward by Aibing Rao and some improved algorithms not only contained the image space information, but also has a good rotation, translation and scale invariance. However, When the color features are extracted, the isolated color blocks with similar color to object has a great influence on the split ring, thereby affecting the results of image retrieval. The improved method is to use (to split rings, which is the average distance between Oi and pixels with same color.(2)Combining the image retrieval algorithm based on bit-plane entropy with the improved image retrieval algorithm based on the circular color histogram, we proposed the image retrieval algorithm based on bit-plane spatial entropy. When the color feature is extracting, the spatial distribution information of each bit-plane is considered. (3)This paper also proposes the image retrieval algorithm based on color vector angle histogram and improved circular color histogram, which neither forget to consider the information of the edge region, nor discard the information contained in the smooth area of the image.Finally, experiments confirmed the validity of those new algorithms.