There ate three problems in the packet loss algorithm of H.264 JM model. Firstly, the actual packet loss rate doesn't often accord with set value. Secondly, it cannot simulate the case of burst packet loss. Finally, if it is experimented with the same stream for many times in the context of fixed packet loss rate, the sequence numbers of lost packets of each experiment are the same. So these problems by calculating the number of lost packets in advance before strict random packet loss are improved, adding Gilbert-Elliott model and evaluating the random number seed with the functions of srand and time. Experiment shows that this algorithm can precisely fulfill the functions of random packet loss and burst packet loss, and at the same time, it can also guarantee the random effect to all experiments.%H.264建议JM模型中的丢包算法存在3个问题:1)实际的丢包率与所设定的预期值经常会有偏差；2)不能模拟突发丢包；3)在丢包率不变的情况下,对同一个码流进行多次丢包实验,则每次实验丢弃包的序号总是一样的.为此,通过预先计算丢包数再严格随机丢包、增加Gilbert-Elliott模型以及利用srand和time函数进行随机数种子赋值对上述问题进行了改进.实验结果表明,改进后的算法能够比较精确地进行随机丢包和突发丢包,并具有随机的效果.
展开▼
