Abstract: Traditional search engine can only search for text data and can not meet the demands of users to gain an image. A search engine of image comparison is presented to establish image feature database by a new approach of fractal image processing and indexing. When user inputs a query image, the search system adopts the same method to get the feature for the query image which is compared with the feature data of image database to realize the expected data searching. Besides, experiment shows that image comparison search engine can get images similar to the query one, it can get right feedback for query image with rotation, faintness, noise and so on. This method has been testified to be error-tolerant and effective.