基于γ-Clahe和Real-esrgan的红外图像增强方法
Method of infrared image enhancement based on γ-Clahe and Real-esrgan
韩龙 1左超 1赵雅婷 1姜楠1
作者信息
- 1. 黑龙江科技大学 电气与控制工程学院,哈尔滨 150022
- 折叠
摘要
针对红外图像对比度低和清晰度差的问题,提出一种基于γ-Clahe和Real-esrgan的红外图像增强方法.通过Haar小波变换将红外图像分解为低频和高频分量,对低频和高频分量分别进行γ-Clahe变换和高斯滤波,将处理后低频和高频分量进行重构得到重构的红外图像,采用Real-esrgan算法对重构的红外图像进行超分辨率重建.结果表明,所提出的红外图像增强算法的主观和客观指标均优于HE、Clahe和Gamma算法,相较于上述三种传统算法 PSNR平均提高了3.525、9.141 和9.631,SSIM平均提高了0.085、0.295 和0.162,使重建后的红外图像对比度和清晰度得到了增强.
Abstract
This paper proposes a infrared image enhancement method based on γ-Clahe and Real-es-rgan to address the low contrast and poor clarity of infrared image.The study involves initially decompo-sing the infrared image into low-frequency and high-frequency components through Haar wavelet trans-form;subjecting the low-frequency componentto γ-Clahe transformation,while filtering the high-frequen-cy component by Gaussian;reconstructing the processed components to generate the enhanced infrared image;and subjecting the enhanced image to super-resolution reconstruction by using Real-esrgan algo-rithm.The experimental results indicate that the proposed method outperforms traditional techniques such as HE,Clahe,and Gamma algorithms in both subjective and objective assessments.Specifically,com-pared to the three traditional methods,the proposed approach yields average improvements in PSNR by 3.525,9.141 and 9.631,and in SSIM by 0.085,0.295 and 0.162,respectively,as which significant-ly enhances the contrast and clarity of the reconstructed infrared images.
关键词
红外图像/Haar小波变换/γ-Clahe/Real-esrganKey words
infrared image/Haar wavelet transform/γ-Clahe/Real-esrgan引用本文复制引用
基金项目
黑龙江省省属高等学校基本科研业务费项目(2022-KYYWF-0527)
黑龙江省省属高等学校基本科研业务费项目(2021-KYYWF-1467)
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据