一种Contourlet图像块编码算法  

A Image Block Coding Algorithm Based on Contourlet

作　　者：张悦庭[1] 孟晓锋[1] 尹忠科[1] 王建英[1] 

机构地区：[1]西南交通大学信息科学与技术学院,四川成都610031

出　　处：《铁道学报》2010年第4期56-62,共7页Journal of the China Railway Society

基　　金：国家自然科学基金资助项目(60772084);四川省应用基础研究计划(2006J13-114)

摘　　要：提出一种高灵活性、低缓冲的Contourlet块编码算法。用临界采样的小波变换取代Contourlet变换中的拉普拉斯塔式LP(Laplacian Pyramid)分解,得到非冗余的Contourlet变换NRCT(Non-Redundant ContourletTransform),使其更适合图像压缩。块编码中,每个子带被划分为相对小的码块,然后对各个码块独立编码,得到具有丰富实用特征集的码流,且明显降低了内存占用。合并NRCT同一小波子带中类似的方向频率划分,根据合并后方向子带的方向频率特性,得到改进的上下文模型。改进的上下文模型使NRCT块编码器中的MQ编码器输出码流码率更加接近其熵率,而JPEG2000上下文模型不能为该MQ编码器输入码流提供良好的条件概率估计。实验证明,特别是对于纹理丰富的图像,本文算法较基于JPEG2000上下文模型的NRCT块编码算法和JPEG2000算法,具有更高的压缩性能。Contourlet block coding scheme is proposed with high flexibility and low memory requirements.A non-redundant version of the Contourlet transform（NRCT） more suitable to apply to image compression is presented by using the critically sampled wavelet transform at the first stage and applying directional filter banks to the wavelet coefficients.Since the block encoder partitions each subband into relatively small code-blocks and codes each of them independently,it produces bit-streams with a rich set of features at an implementation level and requires reduced memory.Assembling the similar directional frequency decompositions in each wavelet subband of NRCT,a modified context model is obtained depending upon the property of the directional frequency in the assembled directional subbands.Benefited from the modified significance context model,the MQ encoder in the NRCT encoder provides the output where bit-rates approximate their entropy-rates more efficiently,while the JPEG2000 context model offers poor initial probability estimation for the input of the MQ encoder.Results show the proposed approach outperforms the NRCT block coding method based on the JPEG2000 context model and the JPEG2000 method,especially for the images that possess abundant textures.

关 键 词：图像压缩 小波 CONTOURLET 上下文 块编码 

分 类 号：TN919.81[电子电信—通信与信息系统]