Gamma变换与多尺度细节增强的红外图像增强  

Infrared image enhancement based on Gamma transformation and multi-scale detail enhancement

作　　者：罗国强 LUO Guo-qiang(School of Information Engineering,Guangzhou Vocational and Technical University of Science and Technology,Guangzhou 510550,China)

机构地区：[1]广州科技职业技术大学信息工程学院,广东广州510550

出　　处：《激光与红外》2023年第2期253-260,共8页Laser & Infrared

基　　金：广东省普通高校特色创新项目(No.2021KTSCX179);广州科技职业技术大学科研基金项目(No.2021ZR05)资助。

摘　　要：针对现有的红外图像增强方法存在欠增强、过增强以及微小细节不清晰等缺陷,提出了一种Gamma变换与多尺度细节增强的红外图像增强方法。该方法充分利用红外图像边缘细节的多尺度特性,用多尺度的引导滤波将红外图像分解为多尺度的细节层和基础层;对基础层进行自适应的Gamma变换,Gamma指数自适应于基础层图像亮区像素数与暗区像素数的比值,旨在适宜地调整红外图像的整体亮度,使得隐藏于暗区的图像信息得以正常显示;对多尺度的细节层分别进行四邻域的拉普拉斯增强,以提升多尺度细节信息的信号强度。实验数据证明,相对于部分最新的现有方法,本文方法具有更好的增强效果,增强图像的视觉感知效果更佳,客观的图像质量指标平均梯度和无参考图像质量指标均较高。Aiming at the defects of existing infrared image enhancement methods, such as under enhancement, over enhancement and blurring of fine details, an infrared image enhancement method based on Gamma transformation and multi-scale detail enhancement is proposed.This method takes full advantage of the multi-scale characteristic of the edges and details of infrared image, decomposes the infrared image into the multi-scale detail layers and the basic layer using multi-scale guided filtering;an adaptive Gamma transformationperforms on the basic layer, the Gamma index is adaptive to the ratio of the number of pixels in the bright area to that in the dark area of the basic layer image, aiming at appropriately adjusting the overall brightness of the infrared image, so that the information hidden in the dark area can be displayed normally;four neighborhood Laplacian enhancement is applied to the multi-scale detail layers to improve the signal strength of the multi-scale detail information.The experimental data provesthe proposed method has a better enhancement effectcompared with some of the latest existing methods, with better visual perception of the enhanced image, and higher objective image quality indexes in terms of average gradient and non-reference image quality index.

关 键 词：图像增强 Gamma变换 多尺度细节增强 引导滤波 平均梯度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]