Compression of Cartoon Images

This book describes a new technique for compressing cartoon images by taking advantage of the distinct color regions in an image and applying a spatially-based compression algorithm. The method for storing these regions involves a combination of Binary Run-Length Encoding and Huffman Encoding. The compression of cartoon images presented here is a lossy compression scheme that removes artifacts and antialiasing before encoding the image, and upon decoding the image, uses an edge-restricted blur filter in an attempt to smooth the edges of the decoded image. Algorithms to support gradient detection, their application, and storage are also described. With these algorithms and the proposed file type, on average the test images were 13.75 times more compact than the corresponding PNG file and 7.45 times more compact than the corresponding JPEG file, with a best bit per pixel ratio of 0.00987 bpp.