Real-Time and Robust Background Updating for Video Surveillance and Monitoring

Abstract

Abstract. Background updating is an important aspect of dynamic scene analysis. Two critical problems: sudden camera perturbation and the sleeping person problem, which arise frequently in real-world surveillance and monitoring systems, are addressed in the proposed scheme. The paper presents a multi-color model where multiple color clusters are used to represent the background at each pixel location. In the proposed background updating scheme, the updates to the mean and variance of each color cluster at each pixel location incorporate the most recently observed color values. Each cluster is assigned a weight which measures the time duration and temporal recurrence frequency of the cluster. The sleeping person problem is tackled by virtue of the observation that at a given pixel location, the time durations and recurrence frequencies of the color clusters representing temporarily static objects are smaller compared to those of color clusters representing the true background colors when measured over a sufficiently long history. The camera perturbation problem is solved using a fast camera motion detection algorithm, allowing the current background image to be registered with the background model maintained in memory. The background updating scheme is shown to be robust even when the scene is very busy and also computationally efficient, making it suitable for real-time surveillance and monitoring systems. Experimental results on real traffic monitoring and surveillance videos are presented.

Detection of stationary foreground objects: A survey

“…In [47], a clustering modeling using color and motion information is proposed, which is robust to lighting changes. In [44], the CM described in [125] is used. The SFO detection strategy in [75] proposes a BG modeling using a modified codebook (the clusters are formed by codewords including appearance information of the pixels) with two layers for detecting, respectively, MOs and SFOs.…”

A Multiscale Parametric Background Model for Stationary Foreground Object Detection

“…The Multiple Gaussian Mixture (MGM) background model [11] addresses this problem by modeling the color distribution at each pixel by a set of Gaussian distributions. The MGM background model and its computationally efficient variants [1], [2], [5], [7] have been observed to perform well under varying illumination conditions and in both, indoor and outdoor environments.…”

“…The version of the MGM background model used in this paper is the one proposed in [1], [7]. The MGM background model incorporates k Gaussian distributions for each pixel of the image I x,y .…”

Light-weight salient foreground detection for embedded smart cameras

“…However, a few Gaussian distributions are usually not sufficient to accurately model backgrounds having fast variations. Methods have been introduced later that are based on Gaussian mixtures [2,14,18,31]. Zivkovic [31] proposed an improved adaptive MoG model to constantly update the parameters of a Gaussian mixture and to simultaneously select the appropriate number of components for each pixel.…”