基于格林函数的图像增强算法  

Image enhancement algorithm based on Green function

作　　者：席志红[1] 赵蓝飞[1] 邢代玉[1] 

机构地区：[1]哈尔滨工程大学信息与通信工程学院,哈尔滨150001

出　　处：《计算机工程与应用》2013年第15期204-209,共6页Computer Engineering and Applications

基　　金：国家自然科学基金(No.60875025)

摘　　要：由于传统的图像增强算法得到的增强图像存在细节的缺失,主观效果较差等缺陷,提出了一种图像增强算法。通过Retinex模型保证了增强图像具有较突出的细节特性;通过求解泊松方程满足了增强图像与原始图像在梯度域的一致性;采用自适应亮度映射得到适于显示的边界条件;对区域的边界进行采样降低算法的复杂度。实验对比了几种图像增强算法得到的增强图像以及相关的评价系数,验证了该算法能够有效地提高图像的对比度,增强图像的主观视觉效果较高。As the traditional image enhancement is suffered from uncomfortable details and visual effect, this paper proposes a new algorithm. The details are ensured by using Retinex model. The consistency in gradient domain between original image and restoration is guaranteed by solving the Poission equation. It adopts the adaptive luminance method to obtain the Dirichlet boundary condition. To reduce the complexity, boundary sampling method is executed. By making comparison with restorations, it shows that this algorithm effectively improves contrast, meanwhile the restorations represent excellent effect of sense.

关 键 词：图像增强 RETINEX 泊松方程 格林函数 梯度域 电像法 狄利克雷边界条件 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]