基于梯度场均衡化的图像对比度增强  

Image Contrast Enhancement by Gradient Field Equalization

作　　者：朱立新[1] 王平安[2] 夏德深[1] 

机构地区：[1]南京理工大学计算机科学与技术学院,南京210094 [2]香港中文大学计算机科学与工程学系

出　　处：《计算机辅助设计与图形学学报》2007年第12期1546-1552,共7页Journal of Computer-Aided Design & Computer Graphics

基　　金：香港特区政府研究资助局资助项目(CUHK/1/00C);江苏省科技厅基金(BS2003655)

摘　　要：在偏微分方程理论框架下提出一种能够有效增强图像中阴影或高亮区域信息的方法.首先对图像梯度场进行直方图均衡化,使这些图像阴影或高亮区域中的细节能够在梯度域得到增强;然后利用最小二乘原理重建出增强后的结果图像.通过引入Lab彩色空间将对比度增强方法推广到对彩色图像的处理中.在数值求解方面,根据Laplacian算子的特点改进了求解Poisson方程的快速算法,改进后的算法具有程序设计简单、计算量小的特点.实验结果表明,文中方法能够有效地改善由于光照影响造成的图像对比度下降.An effective image contrast enhancement method based on partial differential equations is proposed to improve the image quality degraded by the uneven distributed illumination in the image. In order to amplify the details, we adjust the distribution of image gradient by histogram equalization of gradient field firstly and then reconstruct the result image from modified gradient field in least square sense.With the introduction of Lab color space, the new method is extended to color image enhancement. In terms of numerical scheme for Poisson equation, a more efficient scheme is proposed according to the character of the Laplacian operator, which avails the programming and computational efficiency. Experimental results show the effect and efficiency of our new model in improving the image contrast caused by uneven illumination.

关 键 词：偏微分方程 图像增强 直方图均衡化 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]