鲁棒性压缩感知重构技术及其在智能视频监控中的应用研究

Robust compressed sensing reconstruction technique and its application in intelligent video monitoring

在视频监控中需要进行鲁棒性压缩感知重构,降低视频丢包和时延等因素的影响,提出一种基于隐马尔科夫模型的鲁棒性压缩感知重构技术并应用在智能视频监控中。首先利用隐马尔科夫模型对智能视频监控系统中的视频帧序列进行频域特征点奇偶分裂处理;然后计算视频监控编码的标量量化码率分配系数,建立视频压缩感知重构的时空方向树,采用隐马尔科夫模型进行解码重建,实现视频帧的鲁棒性压缩感知重构;最后进行仿真测试。结果表明,采用该方法进行监控视频的压缩感知重构,能有效降低丢包率和传输时延,在智能视频监控中具有较好的应用价值。

As the robustness compressed sensing reconstruction is required in the video monitoring to reduce the effect ofvideo packet loss,time delay and other factors,a robustness compressed sensing reconstruction technology based on hidden Mar?kov model is proposed,which is applied to the intelligent video monitoring.The hidden Markov model is used to deal with odd?even splitting processing of feature points in frequency domain for sequence of video frames in intelligent video surveillance sys?tem,and then the scalar quantization code rate allocation coefficient of the video monitoring encoding is calculated to establish atime?space direction tree for video compressed sensing reconstruction,conduct decoding reconstruction with hidden Markov model,and achieve the robustness compressed sensing reconstruction for video frames.The simulation test results show that the pro?posed method can effectively reduce the packet loss rate and transmission time delay,and has a good application value in intelli?gent video surveillance.