以H.264算法为平台的高效率图像增强算法  

High efficiency image enhancement algorithm for H.264

作　　者：王熹微[1] 刘亚莉[1] 崔慧娟[1] 唐昆[1] 

机构地区：[1]清华大学电子工程系

出　　处：《清华大学学报（自然科学版）》2005年第7期904-907,共4页Journal of Tsinghua University(Science and Technology)

基　　金：国家"九七三"基础研究项目(G1998030406);国家自然科学基金资助项目(60272020)

摘　　要：为了在提高重建图像质量的同时不增加系统开销,提出一种将改进的图像平滑与加权零块判决准则结合的算法。图像增强能有效去除在获取原始图像阶段引入的随机脉冲噪声、Gauss噪声及其混合噪声,但增加了系统开销,视频编码巨大的运算量往往成为实时应用的瓶颈之一。该算法利用改进的MTM(modifiedtrimmedmean)方法提高图像质量,利用加权零块判决准则保持系统性能。以H.264算法为仿真平台进行测试。结果表明:中低码率时,在不增加系统计算开销的前提下,获得同等质量图像,码率可降低50%。Image enhancement for real-time video coding systems effectively removes noise effects due to impulse noise and Gaussian noise from the image capturing stage but significantly increases the algorithm complexity, which limits its real time application. An efficient algorithm which combines image smoothing with the weighted zero-block detection criterion was developed to enhance the quality of reconstructed images without increasing the computational cost. The algorithm exploits the modified MTM （modified trimmed mean） method to improve the quality of decoded images in H.264. The weighted zero-block detection criterion is used to maintain system performance by increasing the number of all-zero blocks after the motion search. Simulations show that in real H.264 applications, the algorithm consistently outperforms the JM61e video coder with more than 50% savings in bitrate without adding excess complexity, especially for low bitrate coding.

关 键 词：视频编码 图像增强 加权零块判决 编码效率 

分 类 号：TN919.81[电子电信—通信与信息系统]