基于局部梯度场均衡化的图像增强方法  

An image enhancement method based on local gradient field equalization

作　　者：董丽丽[1] 丁畅[1] 许文海[1] 

机构地区：[1]大连海事大学信息科学技术学院,辽宁大连116026

出　　处：《光电子．激光》2016年第5期557-565,共9页Journal of Optoelectronics·Laser

基　　金：国家"863"高技术研究发展计划(2014AA0565);国家科技支撑计划(2014BAB12B03);中央高校基本科研业务费专项(3132014301)资助项目

摘　　要：针对低照度环境下图像细节模糊的特点,在偏微分方程理论框架下提出一种有效增强图像阴暗区域中细节信息的方法。首先将图像梯度场压缩到合适的范围内;然后在局部范围内做直方图均衡化处理来建立目标梯度场,最后根据目标梯度场重建出增强后的图像。均衡化梯度场可以使梯度在不同的灰度级上都有体现达到细节增强的目的,对梯度场进行局部均衡化可解决因全局直方图均衡化造成的高亮区域过度扩散的问题。在图像重建过程中,采用矩阵变换法代替Kronecker直积运算求解Poisson方程,有效地降低了数据的存储量,减少程序运行时间。本文算法与传统的直积算法相比,在算法的时间复杂度和空间复杂度上都有很大程度上的改进。实验结果表明,本文算法能够很好地增强图像阴暗区域中的细节信息,尤其是对于极低光照强度下的图像增强。According to the problems of fuzzy details in lower illumination, an effective image enhancement method based on partial differential equation is proposed to improve the dark regionrs quality. Firstly, we compress the dynamical range of image＇s gradient field to the suitable dynamical range. Secondly, we equalize the histogram of the partial range gradient to establish the targeted gradient field. Lastly,we reconstruct the image according to targeted gradient field. Equalizing the gradient field can make gradient amplitude manifest in different grayscales, which can enhance the detailed information. Equalizing the gradient field in partial range can solve the problem of highlighted regionrs diffusion by global equalization of image gradient field. On the process of reconstruction,we use the method of matrix transformation instead of Kronecker＇s calculation to solve the Poisson Equation, which can reduce the data storage capacity effectively and cut down the program running time. Compared with the traditional Kronecker＇s calculation, this paperrs algorithm is improved on the time complexity and space complexity to a large extent. Experimental results show that this paper＇s algorithm can enhance the imagers details more effectively,especially in the condition of lower light.

关 键 词：图像增强 梯度场均衡化 偏微分方程 阴暗区域 细节信息 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]