A New Moving Object Detection Method Based on Frame-difference and Background Subtraction

2017 3rd International Conference on Applied Materials and Manufacturing Technology

Guo, Jiajia^1 ; Wang, Junping^1 ; Bai, Ruixue^1 ; Zhang, Yao^1 ; Li, Yong^1

School of Telecommunication Engineering, XIDIAN University, Xian

710071, China^1

关键词: Background subtraction;    Frame differences;    Gaussian mixtures;    Missed detections;    Morphological processing;    Moving object extraction;    Moving-object detection;    Video surveillance;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/242/1/012115/pdf
DOI  :  10.1088/1757-899X/242/1/012115

来源: IOP

【 摘 要 】

Although many methods of moving object detection have been proposed, moving object extraction is still the core in video surveillance. However, with the complex scene in real world, false detection, missed detection and deficiencies resulting from cavities inside the body still exist. In order to solve the problem of incomplete detection for moving objects, a new moving object detection method combined an improved frame-difference and Gaussian mixture background subtraction is proposed in this paper. To make the moving object detection more complete and accurate, the image repair and morphological processing techniques which are spatial compensations are applied in the proposed method. Experimental results show that our method can effectively eliminate ghosts and noise and fill the cavities of the moving object. Compared to other four moving object detection methods which are GMM, VIBE, frame-difference and a literature's method, the proposed method improve the efficiency and accuracy of the detection.