雾霾天气下以电子眼为基础的户外监控系统几乎全部“瘫痪”——获得的画面模糊不清,无法对监控区域实施有效的监控.在现有的去雾算法中,暗原色先验算法去雾效果最为理想,同时也有一些缺陷和不足,现有的主要问题有:处理时间过长难以达到实时去雾的要求,对天空等明亮区域失效.针对这些问题提出新算法,利用快速导向滤波代替软抠图并加入“容差”机制来改善这些问题,在此基础上,利用校园环境中监控背景几乎不变的特点,提出了加入“收缩-跳跃”机制即间隔性计算图像缩小后的透射率,再将该透射率恢复为原图大小,并利用该透射率对同一间内的其他图像进行去雾的算法,在保证去雾后不影响视频信息提取的前提下,尽可能缩短隔视频处理时间.实验结果表明,运用此方法可以达到校园监控视频去雾的要求.%Electronic eye monitoring system based almost entirely outdoors "paralysis" under fog and haze-get the picture blurred,unable to implement effective monitoring of the monitoring area.In the existing algorithm to fog,dark channel prior to the algorithm fog was most satisfactory,but there are some flaws and shortcomings:long processing time is difficult to achieve real-time defogging and at the request of other bright sky area are useless.To deal with these problems,the proposed algorithm uses fast guided filtering instead of soft matting and adding "tolerance" mechanisms to improve these problems.On this basis,using the characteristic of campus environment is almost constant background monitoring offering a method of adding a "shrinkage-jump mechanism",that is,an interval-calculating image having a reduced transmittance,and then reducing the transmittance to the original image size,and using the transmittance to de-fog the other images in the same cycle.To ensure that after the fog does not affect the extraction of video information under the premise of the video processing time as short as possible.Experimental results show that the use of this method can achieve the requirements of the campus to monitor video fog.
展开▼
