A Compression Domain Image Retrieval Method Using Edge Information And Projection Histogram

Abstract

With the development of multimedia technologies, the problem of how to retrieve a specified image from a large image database becomes an important issue. In this paper, we extract the features directly from the DCT coefficients in the compressed domain as the keys for the retrieval instead of fully reconstructing the compressed images. The advantage is that the complexity and speed of building indices can be efficiently improved. The edge types are used as an image feature for our image retrieval. In the proposed method, users can use the partial images and sketch images to retrieve the desired images. Also, we propose an efficient image retrieval scheme using the projection histogram of edge information to speed up the image retrieving. According to the experiment results, we know that our method is useful for the image retrieval.