An image retrieval algorithm based on improved color histogram

Abstract

Nowadays, with the rise of the Internet, digital images have been developing quickly. The demand of multimedia based on digital image is more and more urgent. Image retrieval algorithms are an effective and accurate algorithm, which search images in large image database. It is widely used in Internet platform and image database. However, most of the proposed methods are based on text content, which can not accurately retrieve the target image according to the visual features and the needs of users. So we proposed an image retrieval algorithm based on improved color histogram, which extracts features based on HSV non-uniform quantized color. Our approach divided the image into several blocks, and calculate the color features of these regions respectively with different weights to search images with each block of color histogram. Our method makes full use of the advantages of image segmentation and HSV inhomogeneity to extract color features for image retrieval. Compared with existing methods, our approach can achieve a higher accuracy.