一种基于感兴趣区的小波图像编码算法  

A New Wavelet Image Coding Algorithm Based on Regions of Interest

作　　者：张立保[1] 王珂[1] 

机构地区：[1]吉林大学通信工程学院,长春130025

出　　处：《计算机工程与应用》2003年第4期22-25,共4页Computer Engineering and Applications

基　　金：国家自然科学基金"城市交通流诱导系统理论模型和方法的研究"的子课题"交通流中的高速图象压缩传输"的资助(编号:59638220)

摘　　要：该文从图像有损压缩后产生的差值图像入手,分析了不同压缩比下嵌入式零树编码对有损压缩图像与原图像的差值图像的影响,结合网络的互动特点,给出了一种基于图像感兴趣区残差图像的小波编码算法。该算法基于图像感兴趣区域的近无损压缩理论,利用了残差图像中的象素呈高斯分布特点。先对小波图像在较低信噪比下做嵌入式零树编码并重建有损图像,求出该图像与原图的差值;然后对该差值图像中的感兴趣区域做DCT编码,将二次编码的数据熵编码后跟在一次编码数据后面传输或存储,从而实现了基于感兴趣区的小波图像编码算法,实验结果表明,该算法不仅提高了压缩效率,而且满足了对感兴趣区图像质量的要求,其简便灵活的特点可以做为小波零树编码的重要补充。From the residual image that is made by the lossy compression,this paper analyzes the affection of different compression ratios based on Embedded Zerotree Wavelet Coding for residual image and net character.And it gives a new wavelet image coding compression algorithm based on residual image distributing characteristic on regions of interest(ROI ).It bases on theory of the near-lossess compression of image ROI and take full advantage of the pixels Gauss distribution character of the residual image.First,the lossy image with low PSNR set up from the wavelet image and the residual image between lossy image and original image is computed.Second,the residual image of ROI is coded by DCT.The coding data is transmitted or stored after the coding rata of lossy image.Experimental results represent that the algorithm can not only obtain much higher compression ratio than lossy wavelet image zero-tree coding compression algorithm,but also can suffice the acquire for the image quantity of region of interest(ROI ).So this algorithm will act the important complement of wavelet zero -tree coding compression because its simple,convenient and flexible characteristic.

关 键 词：小波 图像编码算法 感兴趣区 静止图像压缩 国际标准 图像处理 

分 类 号：TN919.81[电子电信—通信与信息系统] TP391.41[电子电信—信息与通信工程]