Color Facial Image Coding Method Based On Wavelet Transform

Region-Of-Interest (ROI) image coding is a new feature in the newly published JPEG2000 still image compression standard. Low compression ratio compression is used for the region of ROI but high compression ratio compression is used for other regions. Lower compression ratio and better image quality can be acquired by using this algorithm. Because of the greatest data amount, color facial image can't effectively accomplish face recognition under extreme low bit ratio. Due to the importance of facial image in special application, this paper which focused on studying color facial image coding algorithm can choose ROI automatically. Low compression ratio compression is used for facial region but high compression ratio compression is used for other regions. Using the algorithm, we can not only control data quantity of image but guarantee the image quality of the facial feature region. Compared to the traditional algorithm, our algorithm is simple and combines with face detection. Experiment proved that the proposed color facial image coding algorithm based on wavelet can get higher compression ratio while guarantee image quality of facial region. In this paper, the following aspects are emphasized: 1) At first, this paper compares a variety of compression algorithms, including traditional ones and modern ones. It also reviews the up-to-date standards for image compressing. Through analysis and comparison, we choose the most promising wavelet coding algorithm as our basic tool to compress facial images. 2) Through analysis of wavelet theory, it leads wavelet analysis models of digital image. According to the character of transform coefficients, it probes into the principle of the algorithm of wavelet image compressing. 3) We give a promethean coding method based on the embedded coding theory: a ROI coding algorithm in color image based on EZW. Wavelet transform is first step, after that,it chooses the coefficients of ROI according to ROI detection algorithm. By placing a higher emphasis on the transform coefficients pertaining to the ROI, the ROI is coded with higher fidelity than the rest of the image (BG) in earlier stages of progressive reconstruction, thus the "important" part of the image is reconstructed more quickly than BG. 4) We have also discussed a self-adaptation method for face detection based on skin color and template matching. Combining the proposed color image ROI algorithm based on EZW, we get the color facial image coding algorithm based on wavelet.