基于暗元先验图像去雾速度优化技术

摘要

As regard to the time-consuming problem of repeating complex calculation in the process of recovering the picture and getting the neighborhood minimum for pixel in Dark Channel Prior,this paper presents a new method,that the process of getting neighborhood minimum is divided into raw processing and column processing,and the queue is used to save the intermediate value in the process of getting neighborhood minimum,then,a table for defogging is made to avoid the repeated calculation.Experiment results show that the visibility is as good as the original algorithm,and the processing speed is improved greatly,it meets the requirement of the real-time processing.%针对暗元法中邻域最小值求取及图像恢复过程中的重复、复杂计算耗时较大的问题,提出将求取邻域矩形区域最小值过程分为行处理与列处理,并利用队列保存中间量,减少比较次数;制定图像恢复用表避免重复计算等优化技术提高算法速度,经测试证明优化后算法处理结果和原始暗元法去雾效果基本一致,速度有了明显提升,达到实时处理要求.