融合颜色恒常性和多尺度小波的弱光照图像细节增强  

Enhancement of weak illumination image details by fusion of color constancy and multi scale wavelet

作　　者：刘巧玲[1] 刘玲 喻娜[1] LIU Qiaoling;LIU Ling;YU Na(Chengdu University,Chengdu 610106,China;Southwest Minzu University,Chengdu 610000,China)

机构地区：[1]成都大学,成都610106 [2]西南民族大学,成都610000

出　　处：《激光杂志》2025年第4期115-120,共6页Laser Journal

基　　金：国家自然科学基金(No.62302407)。

摘　　要：光照不足易导致图像细节丢失,影响图像的检测和分析。为此,提出一种融合颜色恒常性与多尺度小波技术的弱光照图像细节增强方法。利用B样条曲线精准校正色彩偏差,通过四叉树空间索引与暗通道先验技术实现场景深度重构。应用大气散射模型来复原低频无雾图像,并通过梯度增强提升去雾图像的纹理细节。引入多尺度小波重构技术生成清晰的图像。从输出图像中提取光照权重矩阵,并利用该矩阵完成光照分量的初始估计。在光照结构的约束下,优化估计初始光照分量。通过非线性光照调整进一步优化光照效果,结合Retinex模型增强弱光照图像的细节表现力。实验结果表明,所提方法增强处理后的弱光照图像细节方面更加丰富且清晰。Insufficient lighting can easily lead to loss of image details,affecting image detection and analysis.To this end,a method for enhancing weak lighting image details is proposed,which combines color constancy with multiscale wavelet technology.Using B-spline curves to accurately correct color deviation,scene depth reconstruction is achieved through quadtree spatial indexing and dark channel prior technology.Applying atmospheric scattering models to restore low-frequency fogless images and enhancing the texture details of dehazing images through gradient enhancement.Introducing multi-scale wavelet reconstruction technology to generate clear images.Extract the lighting weight matrix from the output image and use this matrix to complete the initial estimation of lighting components.Under the constraint of lighting structure,optimize the estimation of initial lighting components.Further optimize the lighting effect through nonlinear lighting adjustment,and enhance the detail representation of weak lighting images by combining Retinex model.The experimental results show that the proposed method enhances the details of the processed weak light images to be more abundant and clear.

关 键 词：融合颜色恒常性 多尺度小波 弱光照图像 细节增强 

分 类 号：TN911.73[电子电信—通信与信息系统]