Abstract:
Reference1 proposed a new method based on color features and image bitmap.In this paper, three improvements are proposed as follows: To adopt HSV color space which is more perceptually relevant than RGB color space;To partition an image into equal-sized sub-images to model the color spatial information;By the extraction of Upper Mean color and Lower Mean color, multi-level bitmaps are obtained, which extend the use of comparative information for image differences.In the best case, the performance is improved by an average of 12.16% compared to the original algorithm.