压缩图像码流的分组丢失顽健可伸缩认证算法  

Packet-loss robust scalable authentication algorithm for compressed image streaming

作　　者：易小伟[1,2] 马恒太[1] 郑刚[1] 郑昌文[1] 

机构地区：[1]中国科学院软件研究所天基综合信息系统重点实验室,北京100190 [2]中国科学院大学,北京100049

出　　处：《通信学报》2014年第4期174-181,共8页Journal on Communications

基　　金：国家高技术研究发展计划("863"计划)基金资助项目(2012AA011206)~~

摘　　要：基于图像编码流的结构和相关性特点,提出了一种分组丢失顽健的可伸缩流认证方法。通过利用散列链和纠错编码算法构造认证算法,该方法可实现优化的码率分配以及非平等认证保护(UAP,unequal authentication protection)。首先对图像编码码流进行解析,获得层次结构信息和编解码依赖性;然后,根据码流数据对重构图像质量的重要程度,利用散列链将次重要的码流数据链接到重要数据上;最后对解码独立码流的散列值和整个码流的数字签名进行纠错编码,提高认证算法对分组丢失的顽健性。该方法仅需要对整个图像码流做一次签名,具有很低的认证代价。实验结果表明,与其他3种流认证算法相比,此法的认证图像具有更高的重构质量。Based on structures and dependencies of the image codestreams, a scalable stream-level authentication ap-proach was proposed to resist packet loss. To construct the authentication algorithm by hash chaining and error-correction coding, the proposed approach can realize optimizing bit-rate allocations and unequal authentication protection. Firstly, the compressed streams of the original image are analyzed to obtain hierarchical structures and coding dependencies. Secondly, in accordance with the differentiation-importance of codestreams to the reconstructed image, sub-important packet is linked to more important packets via hash chains. Finally, these hash values of decoding-independent packets and the digital signature of the whole bitstream are encoded with an error-correction coding algorithm. The proposed scheme has a very low authentication overhead because it signs on the whole image once. Experimental results show that the authenticated image of the proposed scheme has high reconstructed quality than the other three stream-level authenti-cation schemes.

关 键 词：图像认证 顽健性 图像编码 流认证 端到端质量 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]