基于暗通道先验和灰度世界的自动白平衡算法改进  

An Improved Automatic White Balance Method Based on Dark Channel Prior Theory and Gray World Assumption

作　　者：李晓川 黄成强[3] 祝永新 田犁[2] 闵嘉华[1] 汪辉[2] 

机构地区：[1]上海大学材料科学与工程学院,上海200444 [2]中国科学院上海高等研究院,上海201210 [3]遵义师范学院,贵州遵义563000

出　　处：《工业控制计算机》2018年第11期90-92,共3页Industrial Control Computer

基　　金：国家重点研发计划(2017YFA206104);上海市科学技术委员会科研计划项目(16511108701);张江管委会公共服务平台项目(2016-14);贵州省科技厅2017年度千层次人才项目(遵市科合人才[2017]13号)

摘　　要：图像传感器在非标准白色色温光源下,拍摄的图像带有色偏,因此需要自动白平衡(AWB)算法。传统白平衡算法在极端光源条件下经常失效。提出了一种更有效的AWB算法。首先,根据暗通道先验理论检测白点区域。之后在白点区域和原图中应用二次映射矩阵。最后,结合非先验灰度世界算法得到精确的转换方程,对原图进行色彩还原。实验研究表明,所提出的方法在各种色温光源下均有显著效果,相比其他方法实现了更好的色彩还原,并具有更高的鲁棒性。Images captured in non-standard white color temperature often contain color cast,so automatic white balance algorithm is very essential.However,classic AWB methods often fail in most extreme light conditions.This paper proposes an efficient AWB method.First,the white point area is detected based on dark channel prior theory.After that,a quadratic trans- mission map is employed in white point area and original image.Finally,non-prior Gray world assumption is applied to gen- erate a refined transmission map.Experimental results indicate that the proposed method is efficient and robust over a wide range of light sources with different color temperatures and achieve better color rendition compared to some other methods.

关 键 词：自动白平衡 暗通道先验算法 二次映射矩阵 FPGA 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]