一种改进的全局Retinex监控视频图像增强方法

考虑到监控视频每相邻帧图像背景及光照近似,合理运用帧间信息,构建了一种改进的Retinex图像增强算法。对每一帧图像进行不同尺度及参数的高斯低通滤波,滤波结果取模极小提取各帧背景信息,通过融合多帧图像背景,获取该时刻上视频图像的准确背景,使用该背景进行Retinex算法增强。实验结果表明,算法增强后的图像暗处细节信息更加丰富,获得了更舒适的视觉效果。

An Improved Retinex Image Enhancement Algorithm

Considering the similarity of each adjacent image's background and light in video surveillance,this paper made rational use of the information between frames and built an improved Retinex image enhancement algorithm.The Gaussian low-pass filter of different scales and parameters for each frame of image,the minimum of the filter's result is gotten,and each frame background's information is abstracted,by merging multi frames of image's background,the video image's accurate background at the moment is obtained and the background to enhance the Retinex algorithm is used.The experimental results show that the details of the image's darkness after the algorithm are richer and obtain more comfortable visual effects.