基于改进的MSR算法在软件视频会议系统上的应用  

An application of an improved MSR algorithm in software video conference system

作　　者：谢小军[1] 孔伟伟[2] 

机构地区：[1]国网安徽省电力公司信息通信分公司,安徽合肥230061 [2]国网电科院安徽南瑞继远软件有限公司,安徽合肥230088

出　　处：《计算机工程与科学》2015年第9期1730-1735,共6页Computer Engineering & Science

摘　　要：视频会议是一种应用广泛的会议方式,由于光照、视频采集设备品质、网络带宽等外界复杂因素的影响,常常导致视频会议视频图像质量下降。在多尺度Retinex滤波(MSR)算法的基础上,使用双边滤波获取原始图像的高低频信息,然后采用颜色约束方法和增益补偿方法增强经过双边滤波处理后的高频反射图像。通过主观体验和客观评价,改进算法增强了视频图像,同时保留了边缘细节,应用于软件视频会议系统,提高了视频图像的清晰度,改善了视觉效果。Video conferences, a new type of conference in the internet age, have been widely used. Due to the influence of the external factors such as light, video acquisition device quality and network bandwidth, the quality of images of video conferences may deteriorate. Based on the multiple scale Ret- inex （MSR） algorithm, we divide the original images into illumination images of low frequency and high frequency reflection images by the bilateral filtering technique. And then the color constraints method and the gain compensation method are adopted to enhance the high frequency reflection images. Subjec- tive experience and objective evaluation show that the improved algorithm enhances video images and preserves edge details when applied to soft video conferences, thus providing a better visual effect.

关 键 词：视频会议 图像增强 MSR 双边滤波 颜色约束 增益补偿 

分 类 号：TN391.4[电子电信—物理电子学]