Image Retrieval Algorithm Based on Fractal Coding

A traditional fractal image retrieval system needs to code all the images before the retrieval, and thus real-time retrieval cannot be realized. Focusing on this problem, this study puts forward an image retrieval algorithm based on image entropy and fractal blocks. In the algorithm, images are screened at first according to comparison of image entropies. Therefore, the screened images in an image library are similar to a query image to some extent. In this way, the number of images requiring to be matched with the query image in the image library can be reduced. In the meanwhile, the time for image retrieval can be greatly shortened. Then, the retrieval function is realized based on the characteristic computation structure similarity of fractal blocks of images. As shown by the experimental results, the algorithm does not need to extract fractal code documents at first when images are put into the library. Thus, defects in offline retrieval of the traditional fractal image retrieval can be overcome. In addition, a precision ratio and a recall ratio of checking results can be ensured.