Improved Compression Algorithm Based on Region of Interest of Face

This paper presents a compression algorithm based on region of interest (ROI) of face to gain a high compressing ratio and to assure the quality of compressed result. First, the ROI including facial features is detected by the FC2VQ (feature correction two-stage vector quantization) algorithm. Then the mask of ROI is calculated. Finally the proposed algorithm is used to compress. The compression ratios of ROI and background can be adjusted randomly according to the requirement. Experimental results show that our algorithm can make the compressing quality of ROI better than that of the background, and the facial features region can be well recognized by eyes