Image Retrieval based on image entropy and blocked matching
Lihong Zhang · Microcomputer Information · 2011
Considering the space distribution feature of color, first we calculate image's overall entropy and make a contrast with set threshold. Based on the theory that similar images have similar image information threshold, we select images similar in general, so pre -classification is realized. Then block irregularly as to these selected images, and work out their gray co -occurrence matrix as their veins feature respectively to do further processing. Then we measure similarity by Euclidean Distance. Experiment shows the way this paper forwarded make the image processing more exact and more quick.