Similarity Detection of Color Image based on Main Color Table

Abstract

With the wider use of huge amount of images, more reasonable retrieval algorithms are becoming more and more important. This paper proposed an image similarity detection algorithm for image content based retrieval. The algorithm can extract the main color of the image, and improve the traditional methods in similarity detection performance. Because the eigenvalues of different regions of the image may be quite different, only using the global features of the image as matching eigenvalues will lead to a large error in similarity judgment, so image segmentation technology has always been the key technology in color image processing. In this paper, a more advanced block technology is used to segment the image of different regions, which can effectively reflect the spatial information of different images.