Abstract: An object-based coding algorithm is proposed for surveillance video compression. Subjective visual quality instead of the traditional PSNR is adopted as the evaluation criterion of reconstructed image so as to fulfill the specific purpose of surveillance video. The surveillance region is enhanced in a controllable manner to improve visual effect, and then lossless compression is conduced. Experimental results show that better subjective quality of surveillance objects region and higher coding efficiency are achieved at the cost of low complexity by the proposed algorithm.

Key words: video surveillance, video compression, image enhancement, lossless coding

摘要： 针对视频监控系统中的视频压缩问题，提出一种基于对象的监控视频编码算法，采用主观视觉质量而不是传统的峰值信噪比作为重建图像评价指标以满足监控视频的特定要求，对监控区域进行可控增强以提高视觉效果。实验结果表明，与H.264相比，该算法能以较低的复杂度实现重建视频中监控对象更高的主观质量，同时具有较高的压缩效率。

关键词: 视频监控, 视频压缩, 图像增强, 无损编码

CLC Number: 

• TP301.6