Formal Photograph Compression in JPEG2000 Coding

A formal facial image compression algorithm is proposed in this paper. Object segmentation techniques is used to locate the facial region firstly. An adapted version of the original Viola-Jones face detector is used to locate eyes and mouth. The position of chin is estimated and then checked according to the curve in the edge image. Secondly, the background is reassigned to decrease unvaluable information in final bitstream. Finally, the facial region is compressed using the region of interest coding technique in JPEG2000. The produced bitstream is totally compatible with the JPEG2000 standard one. As a result, the image compressed by the proposal algorithm could be delivered widely and browsed by many picture tools that support JPEG2000 format. Experiments demonstrate the efficiency of the method.