Open AccessMoving Object Detection Algorithm based on Improved Visual Background Extractor
Author(s) -
Yang xuewen,
Tangyou Liu
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1732/1/012078
Subject(s) - foreground detection , computer science , artificial intelligence , background subtraction , computer vision , robustness (evolution) , pixel , object detection , motion detection , segmentation , background image , jitter , algorithm , pattern recognition (psychology) , motion (physics) , image (mathematics) , telecommunications , biochemistry , chemistry , gene
Aiming at the problems that the ViBe algorithm causes error detection when the camera jitter, ghosts appear when processing moving targets, and the detected moving targets are incomplete, an improved ViBe algorithm based on motion compensation is proposed. In the background modeling stage, the motion compensation method based on KLT is used to obtain background model to enhance the robustness of the algorithm to the dynamic background. In the foreground detection stage, combined the background model of the current pixel with 8 neighbor pixel background model, by the double discriminant algorithm to eliminate ghost areas caused by detecting the real background points as foreground points, introduce OTSU algorithm to obtain the optimal threshold for foreground detection. In the post processing stage, filter the connected components for segmentation masks and update models to limit the wrong background points scattered in the foreground.Finally, the connected domain is repaired by morphological opening operation. The results show that the algorithm can effectively eliminate the interference caused by camera jitter, and suppress the ghost phenomenon to obtain more accurate foreground images.