摘要
用幂次变换函数取代传统的累积积分函数做直方图均衡,使输出图像亮度可调,显示更多细节。对Retinex理论求解法的Kimmel变分模型做了修改,增加了增强图像细节的因式项。将直方图幂次变换函数均衡化法和基于Retinex理论的图像增强法相结合,讨论了结合原则,提出了以图像方差指标作为结合与否的判据,达到了自适应处理薄、浓雾天降质图像的目的。实验表明,该方法既防止了单一进行直方图均衡在图像颜色细节上的畸变,又提高了基于Retinex理论图像增强法去浓雾的能力,取得了较好的视觉效果。
Power transformation function was used in place of traditional accumulation integral for histogram equalization,thus the brightness of output image could be adjusted and more details could be displayed.The Kimmel Retinex algorithm was modified in its variation framework by introducing an expression for protecting the image details.The power transformation function based histogram equalization and the image enhancement method based on Retinex were combined together to adaptively process the fog-degraded images.The principle of combination was discussed,and the variance of image was taken as a criterion of combination.Experiments indicated that this method can not only restrain the aberration of color caused by using only histogram equalization,but also improve the capability of Retinex theory based enhancement method on cleaning dense fog,which may supply a good vision effect.
出处
《电光与控制》
北大核心
2010年第8期52-56,共5页
Electronics Optics & Control
基金
预研项目(513260101)
关键词
图像增强
去雾
RETINEX
幂次变换
直方图均衡
image enhancement
defogging
Retinex
power transformation
histogram equalization
