边缘模型基背景区的JPEG压缩方法研究

Research on Ambient Region of Edge Model Based Compression Using JPEG

模型基编码是目前极富前途的低比特率图象压缩方法之一 ,而边缘模型基方法较之通常的模型基方法有更多的优点 ,正逐渐引起研究人员的注目 .在边缘模型基中 ,除去了图象中大部分高频分量的背景区 ,其灰度变化较缓慢 ,这些被去除了的空域 ,大小大约占整个图象的 10 % .为了能采用传统的 JPEG方法 ,对背景区进行静态压缩 ,提出了一种简易的基于 8× 8子块的直流分量激励区填充方法 .实验表明 ,用该方法得到的填充后的图象十分适合 JPEG压缩 ,且可以得到较高的压缩比 ,同时图象的主观质量也优于直接对未填充图象做 JPEG压缩 .

Model based coding is recognized to be a perspective method of low bit rate image compression method. Due to some merits which cannot be drawn a parallel with ordinary, Edge Model based Coding is given more and more attention by researchers now. In such method, the ambient region of Edge Model based whose most of the higher frequency is cut has a smooth gray change in background. The area which was left is about 10 percent of the whole image. In this paper, in order to compress the ambient image by traditional JPEG method, we present a sensitive region filling method by using 8×8 block filled with DC. Thus the blank area which was filled can provide as much DC as possible. At the receiver, the sensitive region is added to the decompressed ambient region. Experiments and data show that the filled image is adapt to be compressed by using JPEG and can get good compression ratio, simultaneously the subjective quality of reconstructed image is also better than that of by using JPEG directly.