中文摘要：

设计了一个在低码率下，可自动抗噪声的、能有效增强目标区域画质的视频会议系统，系统与H.264标准兼容，采用一种基于彩色肤色模型的方法对初始的人脸进行检测，以获取初始人脸矩形目标区域和背景区域，通过Cam-shift算法对人脸区域进行跟踪，利用一种简单有效的码率控制算法，对不同的区域根据不同的权重分配比特数，从而提高目标区域画质，通过获取相应的量化参数，送入编码器进行编码，同时，系统在初始视频信号输入时，自动对输入视频信号进行噪声检测，在噪声存在的条件下，通过一种新的时空去噪滤波算法，对不同的区域采用不同的去噪算法，提高目标区域的画质，实验测试表明，该系统能够很好地跟踪人脸区域，有效地增强目标区域的画质，同时，该系统能够有效地减弱噪声干扰，保证视频质量。

英文摘要：

A novel video conference system with auto-antinoise characteristic under low bit rate was presented. The system was compatible with H.264 standard. The system detected the initial human face using color skin-stone model in order to obtain the initial rectangle target area of human face and the background area. Then the system tracked down the area of human face using Cam-shift algorithm. System allocated different bit counts for various areas by using a simple but effective rate control algorithm according to their corresponding weights in order to increase the quality of the target area. Then coding was realized by obtaining corresponding quantization parameters. The system automatically detected the noise level of the input video signal and enhanced the quality of the target area by adopting different antinoise ways to different areas through a novel spatio-temporal filter. Experimental results show that the system can correctly track down face region and effectively enhance the quality of target region. The system also can reduce the noise interference and enhance the quality of the region of interest （ROI）.