线性变换与局部均衡融合的红外图像增强  

Infrared image enhancement by fusion of lineartransformation and local equalization

作　　者：魏艳平[1] WEI Yanping(School of Information and Artificial Intelligence,Nanchang Institute of Science and Technology,Nanchang 330108,China)

机构地区：[1]南昌工学院信息与人工智能学院,南昌330108

出　　处：《激光技术》2024年第5期705-710,共6页Laser Technology

基　　金：国家自然科学基金资助项目(61562063);江西省教育厅科学技术研究项目(GJJ2202909,GJJ191092);江西省高等学校教学改革课题(JXJG-21-27-3);南昌工学院科技计划博士专项基金资助项目(NGKJ-22-01)。

摘　　要：为了改善红外图像的效果,提升对比度和清晰度,丰富边缘细节信息,提出了融合线性变换和局部均衡的红外图像增强方法。利用图像的像素值分布,对像素值进行自适应的分段线性变换,并用局部的直方图均衡增强图像;分别计算两张增强图像的权重图。对比度权重、显著性权重和亮度分布权重;以拉普拉斯金字塔和高斯金字塔的方式,分别对增强图像和权重图进行分解,将分解的图像与权重图进行多尺度线性融合,获得效果理想的增强图像。结果表明,相对于现有方法,本文中提出的方法增强图像的视觉效果更清晰,信息熵、平均梯度和变异系数分别比现有方法高出9.03%、23.87%和9.97%以上。该研究可更有效地提高红外图像增强的性能。Aiming at improving the effect of infrared image,increasing its contrast and clarity and enriching its edge and detail information,an infrared image enhancement method by fusion of linear transformation and local equalization was proposed.According to the intensity distribution of image,adaptive piecewise linear transformation on the intensity of pixel was performed,and local histogram equalization on the image was carried out.Then,the contrast weight,saliency weight and brightness weight of the two enhanced images were calculated,respectively.Finally,multi-scale Laplace pyramid decomposition and Gaussian pyramid decomposition were performed on the enhanced images and the corresponding weights,respectively,and multi-scale linear fusion with the decomposed images and the corresponding weights were performed to obtain the final enhanced image.According to the experimental results,it is confirmed that the effectiveness of proposed method compared to existing methods,the enhanced images go with better visual effect,and the information entropy,average gradient and coefficient of variation are higher than existing methods by more than 9.03%,23.87%and 9.97%,respectively.This study could improve infrared image enhancement performance more effectively.

关 键 词：图像处理 对比度 变异系数 像素变换 局部直方图均衡 多尺度金字塔融合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]