首页|面向不均匀光照的非配对低照度彩色图像增强方法

面向不均匀光照的非配对低照度彩色图像增强方法

扫码查看
为了改善不均匀光照环境下低照度彩色图像的增强效果,解决彩色图像增强处理中颜色易失真问题,提高后续图像处理的质量,提出一种面向不均匀光照的低照度彩色图像增强方法.通过构建非配对低照度图像增强框架,创建一种基于多条件融合的深度学习网络模型.利用非配对框架中的渲染网络和增强网络循环执行,分别以深度信息、颜色图和噪声图作为条件引导,最终实现低照度彩色图像的增强处理.与现有方法进行对比,文中方法能够使不均匀光照下的低照度彩色图像在色彩校正和对比度提升方面同时获得理想效果.
An unpaired low light color image enhancement method for uneven illumination
In order to improve the enhancement effect of low light color images in uneven illumination environments,solve the problem of color distortion in color image enhancement processing,and improve the quality of subsequent image processing,this paper proposes a low light color image enhancement method for uneven illumination.Create a deep learning network model based on multi condition fusion by constructing a non paired low light image enhancement framework.By utilizing the rendering network and enhancement network in a non paired framework to loop through,depth information,color maps,and noise maps are used as conditional guidance,ultimately achieving enhancement processing of low light color images.Compared with existing experimental methods,this method can achieve ideal results in both color correction and contrast enhancement for low light color images under uneven illumination.

machine learningdeep learningimage enhancementgenerative adversarial network

张宪红、王期初

展开 >

黑龙江工程学院 计算机科学与技术学院,哈尔滨 150050

黑龙江工程学院 测绘工程学院,哈尔滨 150050

机器学习 深度学习 图像增强 生成对抗网络

2024

黑龙江工程学院学报
黑龙江工程学院

黑龙江工程学院学报

影响因子:0.414
ISSN:1671-4679
年,卷(期):2024.38(5)