基于人眼视觉特性的Curvelet域低照度图像增强  

Low-light image enhancement in Curvelet domain based on human visual characteristics

作　　者：郑云海 郭建钊 王门鸿 李佳 李庆武 ZHENG Yunhai;GUO Jianzhao;WANG Menhong;LI Jia;LI Qingwu(State Grid Quanzhou Electric Power Supply Company,Quanzhou 362000,China;College of Internet of Things Engineering,Hohai University,Changzhou 213022,China)

机构地区：[1]国网泉州供电公司,福建泉州362000 [2]河海大学物联网工程学院,江苏常州213022

出　　处：《应用科技》2018年第6期27-31,共5页Applied Science and Technology

基　　金：国家自然科学基金项目(41706103);国家电网科技项目(52133016000U)

摘　　要：针对低照度下图像降质严重的问题,提出了一种基于人眼视觉特性和Curvelet变换的低照度图像增强算法。首先将低照度图像转换至"色调-饱和度-亮度"(HSI)颜色空间,在Curvelet域中分解亮度参量得到不同尺度、不同方向的子带分量,以此构建人眼视觉模型;然后利用模型的亮度遮蔽特性和亮度-对比度遮蔽特性对高频分量进行非线性增强,同时对低频分量进行非线性拉伸;最后通过Curvelet逆变换重构亮度参量,结合原始图像的色度和饱和度分量将图像转换至原色彩空间,得到增强后的低照度图像。实验结果表明,该算法可以有效提升低照度图像的对比度和亮度,保持图像的细节信息,抑制图像噪声。Aiming at the serious degradation of images under low illumination,this paper proposes a low-light image enhancement method based on human visual characteristics and Curvelet transform.Firstly,transform the low-light image into the hue-saturation-intensity(HSI)space,and obtain the sub-band components of different scales and directions in Curvelet domain,then construct the human visual model on this basis;and further enhance the highfrequency components nonlinearly based on the luminance masking and luminance-contrast masking characteristic of the model,and at the same time,stretch the low-frequency components nonlinearly.Finally,reconstruct the brightness components by the inverse Curvelet transform.After the combination with the hue and saturation components,convert the image to the original color space to obtain the enhanced low-lightness image.Experimental results demonstrate that the algorithm can effectively enhance the contrast and brightness of low-light images,keep the details of images and suppress the noise of images.

关 键 词：图像处理 图像增强 低照度图像 CURVELET HSI空间 人眼视觉特性 亮度遮蔽 亮度-对比度遮蔽 

分 类 号：TP394.1[自动化与计算机技术—计算机应用技术]