Moving Object Detection Based on Non-Convex RPCA With Segmentation Constraint
Recently, robust principal component analysis (RPCA) has been widely used in the detection of moving objects. However, this method fails to effectively utilize the low-rank prior information of the background and the spatiotemporal continuity prior of the moving object, and the target extraction eff...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2020-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9018279/ |
| _version_ | 1828407052135825408 |
|---|---|
| author | Zixuan Hu Yongli Wang Rui Su Xinxin Bian Hongchao Wei Guoping He |
| author_facet | Zixuan Hu Yongli Wang Rui Su Xinxin Bian Hongchao Wei Guoping He |
| author_sort | Zixuan Hu |
| collection | DOAJ |
| description | Recently, robust principal component analysis (RPCA) has been widely used in the detection of moving objects. However, this method fails to effectively utilize the low-rank prior information of the background and the spatiotemporal continuity prior of the moving object, and the target extraction effect is often poor when dealing with large-scale complex scenes. To solve the above problems, a new non-convex rank approximate RPCA model based on segmentation constraint is proposed. Firstly, the model adopts the low-rank sparse decomposition method to divide the original video sequence into three parts: low-rank background, moving foreground and sparse noise. Then, a new non-convex function is proposed to better constrain the low-rank characteristic of the video background. Finally, based on the spatiotemporal continuity of the foreground object, the video is segmented by the super-pixel segmentation technology, so as to realize the constraint of the motion foreground region. The augmented Lagrange multiplier method is used to solve the model. Experimental results show that the proposed model can effectively improve the accuracy of moving object detection, and has better visual effect of foreground object detection than existed methods. |
| first_indexed | 2024-12-10T11:19:58Z |
| format | Article |
| id | doaj.art-42e2064448ea4d7185d2615ae65304ad |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-10T11:19:58Z |
| publishDate | 2020-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-42e2064448ea4d7185d2615ae65304ad2022-12-22T01:51:00ZengIEEEIEEE Access2169-35362020-01-018410264103610.1109/ACCESS.2020.29772739018279Moving Object Detection Based on Non-Convex RPCA With Segmentation ConstraintZixuan Hu0https://orcid.org/0000-0002-8433-3317Yongli Wang1https://orcid.org/0000-0001-8221-451XRui Su2https://orcid.org/0000-0002-3642-8522Xinxin Bian3https://orcid.org/0000-0002-0495-5141Hongchao Wei4https://orcid.org/0000-0001-9374-4751Guoping He5https://orcid.org/0000-0002-7988-3875College of Mathematics and Systems Science, Shandong University of Science and Technology, Qingdao, ChinaCollege of Mathematics and Systems Science, Shandong University of Science and Technology, Qingdao, ChinaQingdao Urban Planning & Design Research Institute, Qingdao, ChinaCollege of Mathematics and Systems Science, Shandong University of Science and Technology, Qingdao, ChinaCollege of Mathematics and Systems Science, Shandong University of Science and Technology, Qingdao, ChinaShandong Computing Center, Qilu University of Technology, Jinan, ChinaRecently, robust principal component analysis (RPCA) has been widely used in the detection of moving objects. However, this method fails to effectively utilize the low-rank prior information of the background and the spatiotemporal continuity prior of the moving object, and the target extraction effect is often poor when dealing with large-scale complex scenes. To solve the above problems, a new non-convex rank approximate RPCA model based on segmentation constraint is proposed. Firstly, the model adopts the low-rank sparse decomposition method to divide the original video sequence into three parts: low-rank background, moving foreground and sparse noise. Then, a new non-convex function is proposed to better constrain the low-rank characteristic of the video background. Finally, based on the spatiotemporal continuity of the foreground object, the video is segmented by the super-pixel segmentation technology, so as to realize the constraint of the motion foreground region. The augmented Lagrange multiplier method is used to solve the model. Experimental results show that the proposed model can effectively improve the accuracy of moving object detection, and has better visual effect of foreground object detection than existed methods.https://ieeexplore.ieee.org/document/9018279/Moving object detectionrobust principal component analysisnon-convex rank approximationvideo segmentation |
| spellingShingle | Zixuan Hu Yongli Wang Rui Su Xinxin Bian Hongchao Wei Guoping He Moving Object Detection Based on Non-Convex RPCA With Segmentation Constraint IEEE Access Moving object detection robust principal component analysis non-convex rank approximation video segmentation |
| title | Moving Object Detection Based on Non-Convex RPCA With Segmentation Constraint |
| title_full | Moving Object Detection Based on Non-Convex RPCA With Segmentation Constraint |
| title_fullStr | Moving Object Detection Based on Non-Convex RPCA With Segmentation Constraint |
| title_full_unstemmed | Moving Object Detection Based on Non-Convex RPCA With Segmentation Constraint |
| title_short | Moving Object Detection Based on Non-Convex RPCA With Segmentation Constraint |
| title_sort | moving object detection based on non convex rpca with segmentation constraint |
| topic | Moving object detection robust principal component analysis non-convex rank approximation video segmentation |
| url | https://ieeexplore.ieee.org/document/9018279/ |
| work_keys_str_mv | AT zixuanhu movingobjectdetectionbasedonnonconvexrpcawithsegmentationconstraint AT yongliwang movingobjectdetectionbasedonnonconvexrpcawithsegmentationconstraint AT ruisu movingobjectdetectionbasedonnonconvexrpcawithsegmentationconstraint AT xinxinbian movingobjectdetectionbasedonnonconvexrpcawithsegmentationconstraint AT hongchaowei movingobjectdetectionbasedonnonconvexrpcawithsegmentationconstraint AT guopinghe movingobjectdetectionbasedonnonconvexrpcawithsegmentationconstraint |