摘要
针对现有暗通道优先(DCP)的图像去雾算法对全局大气光估计不准确和去雾后图像的细节不突出等问题进行了改进研究,提出了一种改进的中值暗通道优先(MDCP)图像去雾算法。该算法首先利用迭代最大类间差法(Otsu算法)锁定近似理想的天空区域,计算全局大气光,然后运用MDCP原理获得随深度变化的大气传输衰减过程,以达到突出细节信息等目的,再运用软消光(soft matting)算法对大气传输衰减过程进行优化,最后得到场景的反射图像,实现图像去雾。通过实验证明,该方法不仅能够很好地锁定背景处的有效区域,准确地估计全局大气光,还能够很好地保留图像细节信息,使图像在去雾后还能够强化细节信息,达到理想的去雾效果。
A new image defog algorithm based on the median dark channel prior（MDCP） was studied to deal with the existing DCP defoggingies methods＇ problems of inaccurate estimation of the global atmospheric light and no distinct image details after image defogging. The new algorithm uses an iterative Otsu algorithm to locate the ideal sky region and calculate the global atmospheric light, and then utilites the MDCP algorithm to get the atmospheric transmission which declines with the changing of the depth of field, and applies the soft matting to optimization of the atmospheric transmission, and finally obtains the reflectance map to realize the effect of image defogging. The experimental results show that the proposed method can not only locate the effective region on the backgrounds very well, estimate the global atmospheric light precisely, but also reserve image details, strength the detail information, and realize the idea effect of defogging.
出处
《高技术通讯》
CAS
CSCD
北大核心
2014年第5期492-497,共6页
Chinese High Technology Letters
基金
国家科技支撑计划(2012BAH31B01)
北京市自然科学基金(B类)(KZ201310028035)资助项目
