基于自适应遗传算法的SVC非均等错误保护算法  

Unequal error protection with adaptive genetic algorithm for scalable video coding

作　　者：田波[1] 杨宜民[1] 蔡述庭[1] 

机构地区：[1]广东工业大学自动化学院,广州510090

出　　处：《计算机应用》2014年第1期162-166,共5页journal of Computer Applications

基　　金：国家自然科学基金资助项目(61201392)

摘　　要：为提高可伸缩视频编码(SVC)在丢包的网络传输环境下的抗误码性能,提出了一种基于自适应遗传算法的SVC非均等错误保护算法。首先针对可伸缩视频编码的网络抽象层单元数据包头的特点,设计了一种新的网络抽象层单元的封装方案。然后将前向纠错编码的校验位在各层的分配转化为多约束条件下的优化问题,再引入惩罚函数将多约束优化问题转化为无约束优化问题,进而采用自适应遗传算法进行求解。仿真实验结果表明,与目前典型的非均等错误保护算法相比,该算法使重建的可伸缩视频编码的峰值信噪比的平均值提高了0.8 dB^1.95 dB,并有效提高了可伸缩视频编码在接收端的解码速度和重建质量。In order to improve the packet loss resilience of Scalable Video Coding （SVC） over communication networks, an efficient Unequal Error Protection （UEP） algorithm for SVC using adaptive genetic algorithm was proposed. A method to encapsulate network abstract layer units according to the feature of the head information of a packet was introduced. Then the problems of pair codes assignment were transformed into the problems of multi-constraint optimization, which could be transformed into unconstrained objective by exploiting penalty function. Therefore, the adaptive genetic algorithm was employed to obtain globally optimal solution. The simulation results reveal that compared with the typical unequal error protection algorithms, the Peak Signal-to-Noise Ratio （PSNR） is improved by 0.8 dB - 1.95 dB, and the proposed algorithm provides substantial improvement for the decoding speed and received video quality over best effort packet networks.

关 键 词：可伸缩视频编码 非均等错误保护 自适应遗传算法 优化 丢包率 

分 类 号：TN919.85[电子电信—通信与信息系统]