基于图像内容和人眼视觉特性的JPEG压缩编码

JPEG Compressed Encoding Based on the Input Image Contents and Property of HVS

静止图像压缩主要包括变换、量化和编码 ,其中量化对压缩的性能 (图质和压缩比 )起决定作用 ,它涉及质量因子和基本量化表的确定。从输入图像的统计特性出发 ,通过比特率控制算法 ,统计图像的活动性水平 ,选择适当的质量因子使码流文件大小不超过给定值。并根据人类视觉系统的特征 ,计算出更能反映人眼特性的基本量化表 ,将二者结合起来用在JPEG基本压缩系统中 ,使系统的整体性能在率失真理论的意义上得到优化。软件验证表明 ,采用此算法进行压缩 ,在保证码流文件大小的同时 。

Among transformation, quantization and encoding steps in still image compression, quantization is the key of the performances(i.e. image quality and compression ratio) of compression, which is related to the determination of quality factor and basic quantization table. In this paper, based on the statistical characteristics of the input image, it is guaranteed that the file size of the compressed code stream is no more than the predefined size with the bit rate control algorithm which is realized through the calculation of the image activity level and appropriate selection of the quality factor. According to the property of HVS(human visual system), the image-specific basic quantization table is worked out. The above two methods are combined and applied to the JPEG baseline compression system, and the overall performance is optimized in the light of rate-distortion theory. Software verification result reveals that with this algorithm, the file size of the compressed image is guaranteed and the visual effect of the decompressed image is good.