Ruyi Han , Shenghai Liao , Shujun Fu , Yuanfeng Zhou , Yuliang Li , Hongbin Han
{"title":"Inter-frame feature fusion enhanced spatio-temporal consistent video inpainting with sample-based techniques and adaptive local search","authors":"Ruyi Han , Shenghai Liao , Shujun Fu , Yuanfeng Zhou , Yuliang Li , Hongbin Han","doi":"10.1016/j.cam.2025.116523","DOIUrl":null,"url":null,"abstract":"<div><div>Video inpainting is an ill-posed inverse problem in image processing focused on eliminating unwanted objects and restoring damaged or corrupted regions within a video sequence. During video transmission, challenges such as network bandwidth limitations and transmission errors can lead to the loss of crucial information in specific frames. To address this issue, we propose a novel and effective sample-based video inpainting method in this paper. Firstly, an extension frame search strategy is introduced, utilizing image fusion techniques to integrate information from extension frames for more effective restoration of damaged areas and ensuring temporal consistency in the repair process. Secondly, a new priority scheme is proposed to accurately measure the inpainting order of image blocks and to effectively inpaint the structure and texture of video frames. An adaptive local search window is then designed to find the optimal matching block for a target block, ensuring spatial continuity in restoration and reducing the time cost of the proposed algorithm. Additionally, to enhance matching accuracy, a Weighted Sum of Squared Differences (WSSD) matching method based on a facet model is presented in this work. Finally, the proposed method is tested on videos with various backgrounds and damaged blocks of different sizes and locations. Experimental results indicate that our method outperforms six state-of-the-art video inpainting methods both qualitatively and quantitatively.</div></div>","PeriodicalId":50226,"journal":{"name":"Journal of Computational and Applied Mathematics","volume":"464 ","pages":"Article 116523"},"PeriodicalIF":2.1000,"publicationDate":"2025-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Computational and Applied Mathematics","FirstCategoryId":"100","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S037704272500038X","RegionNum":2,"RegionCategory":"数学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"MATHEMATICS, APPLIED","Score":null,"Total":0}
引用次数: 0
Abstract
Video inpainting is an ill-posed inverse problem in image processing focused on eliminating unwanted objects and restoring damaged or corrupted regions within a video sequence. During video transmission, challenges such as network bandwidth limitations and transmission errors can lead to the loss of crucial information in specific frames. To address this issue, we propose a novel and effective sample-based video inpainting method in this paper. Firstly, an extension frame search strategy is introduced, utilizing image fusion techniques to integrate information from extension frames for more effective restoration of damaged areas and ensuring temporal consistency in the repair process. Secondly, a new priority scheme is proposed to accurately measure the inpainting order of image blocks and to effectively inpaint the structure and texture of video frames. An adaptive local search window is then designed to find the optimal matching block for a target block, ensuring spatial continuity in restoration and reducing the time cost of the proposed algorithm. Additionally, to enhance matching accuracy, a Weighted Sum of Squared Differences (WSSD) matching method based on a facet model is presented in this work. Finally, the proposed method is tested on videos with various backgrounds and damaged blocks of different sizes and locations. Experimental results indicate that our method outperforms six state-of-the-art video inpainting methods both qualitatively and quantitatively.
期刊介绍:
The Journal of Computational and Applied Mathematics publishes original papers of high scientific value in all areas of computational and applied mathematics. The main interest of the Journal is in papers that describe and analyze new computational techniques for solving scientific or engineering problems. Also the improved analysis, including the effectiveness and applicability, of existing methods and algorithms is of importance. The computational efficiency (e.g. the convergence, stability, accuracy, ...) should be proved and illustrated by nontrivial numerical examples. Papers describing only variants of existing methods, without adding significant new computational properties are not of interest.
The audience consists of: applied mathematicians, numerical analysts, computational scientists and engineers.