基于融合稀疏模型的引导单幅图像去雨算法被引量：1

Guided single image rain removal algorithm based on fusion sparse model

在线阅读下载PDF

导出

摘要针对现有单幅图像去雨算法存在雨线残留较明显或背景过度平滑的问题,本文提出融合稀疏模型的引导单幅图像去雨算法。通过交替方向乘子法分别对融合解析和综合稀疏模型学习出含部分雨线残留的背景图和雨线图;采用具有较好保边平滑性的引导滤波得到最终去雨图。实验结果表明:所提算法在峰值信噪比和结构相似性方面均比近三年典型对比算法有所提高,且有效去除雨线的同时能较大程度地保留图像背景细节信息。 Aiming at the problem that the existing single image rain removal algorithms exist obvious rain streaks residual or over-smooth background,a guided single image rain removal algorithm based on fusion sparse model is proposed. Firstly,the fusion analysis and synthesis sparse model are used to learn the background image with partial rain streaks residual and the rain streaks image by iterative method that called the alternating direction method of multipliers. Secondly,the guided filtering with better edge preservation smoothness is used to obtain the final rain-free image. The experimental results show that the proposed algorithm has higher peak signal to noise ratio and structural similarity than the typical comparison algorithm in the past three years. And the proposed algorithm can remove rain streaks effectively and preserve image background detail to a large extent at the same time.

作者王科平蔡凯利杨少沛杨艺 WANG Keping;CAI Kaili;YANG Shaopei;YANG Yi(College of Electrical Engineering and Automation,Henan Polytechnic University,Jiaozuo 454150,China;Huanghe Jiaotong University,Jiaozuo 454950,China)

机构地区河南理工大学电气工程与自动化学院黄河交通学院

出处《传感器与微系统》 CSCD 2020年第4期148-151,160,共5页 Transducer and Microsystem Technologies

基金国家重点研发计划资助项目(2018YFC0604500) 河南省科技攻关项目(172102210270) 河南省教育厅科学技术研究重点项目(15A510025) 河南省高等学校重点科研项目(17A480007)。

关键词单幅图像去雨解析稀疏模型综合稀疏模型交替方向乘子法引导滤波 single image rain removal analysis sparse model synthesis sparse model alternating direction method of multipliers guided filtering

分类号 TP391 [自动化与计算机技术—计算机应用技术]

作者简介王科平(1976-),女,副教授,主要研究领域为图像处理、目标跟踪;通讯作者:蔡凯利(1991-),女,硕士研究生,研究方向为图像处理。