特征分离和非阴影信息引导的阴影去除网络  

Feature separation and non-shadow information-guided shadow removal network

作　　者：黄颖[1,2] 房少杰 程彬 姜茂 钱鹰[1] HUANG Ying;FANG Shaojie;CHENG Bin;JIANG Mao;QIAN Ying(School of Software Engineering,Chongqing University of Posts and Telecommunications,Chongqing 400065,China;College of Computer Science and Technology,Chongqing University of Posts and Telecommunications,Chongqing 400065,China)

机构地区：[1]重庆邮电大学软件工程学院,重庆400065 [2]重庆邮电大学计算机科学与技术学院,重庆400065

出　　处：《通信学报》2024年第5期178-190,共13页Journal on Communications

基　　金：国家自然科学基金重点项目(No.62331008)。

摘　　要：为了解决现有阴影去除方法中存在的性能瓶颈以及去除结果产生的色差问题,构建了一个特征分离和非阴影信息引导的阴影去除网络(FSNIG-ShadowNet)。在分离和重建阶段,利用自重建监督将阴影图像分离成直射光和环境光两部分,对光照类型和反射率进行特征解耦分离,设计解码器对分离的特征进行重新耦合以获得无阴影图像。在细化阶段,该网络关注阴影和非阴影的邻接区域,设计局部区域自适应归一化模块将局部非阴影区域颜色先验传递至阴影区域以减少两区域间的色差。实验结果表明,所提FSNIG-ShadowNet与其他优秀的方法相比取得了较有竞争力的结果。To tackle the performance bottlenecks and color deviation issues stemming from current shadow removal methods,a feature separation and non-shadow information guided shadow removal network(FSNIG-ShadowNet)was constructed.In the separation and reconstruction stage,the shadow image was separated into direct light and ambient light using self-reconstruction supervision,with decoupling of lighting types and reflectance.Subsequently,a decoder was employed to re-couple the separated features to yield shadow-free images.In the refinement stage,the network fo‐cused on the adjacent regions of shadow and non-shadow,incorporating a local region adaptive normalization module to transfer the color priors of local non-shadow region to shadow regions for mitigating color deviation between the two re‐gions.Experimental results demonstrate that the proposed FSNIG-ShadowNet achieves competitive results compared to other state-of-the-art methods.

关 键 词：阴影去除 特征分离 自重建 颜色先验 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]