Motion Compensation Based Fast Moving Object Detection in Dynamic Background
This paper investigates robust and fast moving object detection in dynamic background. A motion compensation based approach is proposed to maintain an online background model, then the moving objects are detected in a fast fashion. Specifically, the pixel
- PDF / 634,253 Bytes
- 10 Pages / 439.37 x 666.142 pts Page_size
- 78 Downloads / 247 Views
Abstract. This paper investigates robust and fast moving object detection in dynamic background. A motion compensation based approach is proposed to maintain an online background model, then the moving objects are detected in a fast fashion. Specifically, the pixel-level background model is built for each pixel, and is represented by a set of pixel values drawn from its location and neighborhoods. Given the background models of previous frame, the edge-preserving optical flow algorithm is employed to estimate the motion of each pixel, followed by propagating their background models to the current frame. Each pixel can be classified as foreground or background pixel according to the compensated background model. Moreover, the compensated background model is updated online by a fast random algorithm to adapt the variation of background. Extensive experiments on collected challenging videos suggest that our method outperforms other state-of-the-art methods, and achieves 8 fps in efficiency. Keywords: Fast object detection · Random algorithm · Dynamic background · Motion compensation
1
Introduction
Moving object detection with dynamic background is to detect moving objects under a moving camera, and has a broad prospect of application and research value in the intelligent transportation, medical diagnosis, security monitoring, and many other industries. However, due to the high complexity of the existing method which are unable to meet the time demand of many applications, it is still a challenging subject in computer vision. Aimed at overcoming this limitation, this paper proposes a fast moving object detection framework in dynamic background, in which the motion compensation algorithm is utilized to accommodate the dynamic background, and the background model is updated online in a probability way to adapt the variation of background. Specifically, the background model of each pixel consists of a set of pixels, which are initialized by its location and neighbors. When new frame arriving, the optical flow algorithm, based on edge-preserving patch matching c Springer-Verlag Berlin Heidelberg 2015 H. Zha et al. (Eds.): CCCV 2015, Part II, CCIS 547, pp. 247–256, 2015. DOI: 10.1007/978-3-662-48570-5 24
248
W. Zhang et al.
is employed to compensate the motion of each pixel and propagate their background models from previous frame to current one. Then, every pixel can be classified as the foreground or background pixel by the matching score with their background models. Furthermore, the background models are updated in an online fashion to adapt the variation of background. To the best of our knowledge, it’s the first time to develop a near real-time moving object detection in dynamic background. The key contributions of this paper are summed up in three aspects. Firstly, a general framework is proposed for robustly and fast detecting moving objects in dynamic background, in which the detection speed can reach near real-time. Secondly, a robust background model based on motion compensation is developed and updated online by
Data Loading...
