摘要
视频传感器网络在监控过程中会产生大量数据,为此提出一种改进的基于对象的视频压缩方法。首先,运用改进的混合高斯模型进行背景建模;其次,提出宏块分类方法,对视频帧图像进行宏块分类;然后,利用宏块分类结果,实现前景和背景的分离;最后,分别对前景和背景进行压缩,形成各自的编码流。实验结果表明,该方法具有比MPEG-4更强的压缩能力,满足视频传感器网络的视频压缩需求。
In the monitoring process, the video sensor networks will produce large amounts of data. For this problem, this paper presents an improved method for object-based video compression. Firstly, the improved Gaussian mixture model is used to model background. Secondly, macroblock classification method is proposed to classify the macroblock of video frame image. Thirdly, the results of macroblock classification is used to realize the separation between foreground and background. Finally, foreground and background are separately compressed to form a respective encoded streams. Experimental results show that the method is stronger than the MPEG-4 and meet the needs of video compression in video sensor networks.
出处
《计算机时代》
2015年第4期6-8,12,共4页
Computer Era
