Conference Proceeding

Automatic Robust Background Modeling Using Multivariate Non-parametric Kernel Density Estimation for Visual Surveillance.

01/2005; DOI:10.1007/11595755_44 In proceeding of: Advances in Visual Computing, First International Symposium, ISVC 2005, Lake Tahoe, NV, USA, December 5-7, 2005, Proceedings
Source: DBLP

ABSTRACT The final goal for many visual surveillance systems is auto- matic understanding of events in a site. Higher level processing on video data requires certain lower level vision tasks to be performed. One of these tasks is the segmentation of video data into regions that corre- spond to objects in the scene. Issues such as automation, noise robust- ness, adaptation, and accuracy of the model must be addressed. Current background modeling techniques use heuristics to build a representation of the background, while it would be desirable to obtain the background model automatically. In order to increase the accuracy of modeling it needs to adapt to different parts of the same scene and finally the model has to be robust to noise. The building block of the model representation used in this paper is multivariate non-parametric kernel density estima- tion which builds a statistical model for the background of the video scene based on the probability density function of its pixels. A post pro- cessing step is applied to the background model to achieve the spatial consistency of the foreground objects.

0 0
  • Source
    [show abstract] [hide abstract]
    ABSTRACT: Video segmentation is one of the most important tasks in high-level video processing applications. Stationary cameras are usually used in applications such as video surveil- lance and human activity recognition. However, possible changes in the background of the video such as waving ∞ags, ∞uctuating monitors, water surfaces, etc. make the detec- tion of objects of interest particularly challenging. These types of backgrounds are called quasi-stationary backgrounds. In this paper we propose a novel, non-statistical technique to generate a background model and use this model for background subtraction and fore- ground region detection in the presence of such challenges. The main advantage of the proposed method over the state of the art is that unlike statistical techniques the accu- racy of foreground regions is not limited to the estimate of the probability density. Also, the memory requirements of our method are independent of the number of training sam- ples. This makes the proposed method useful in various scenarios including the presence of slow changes in the background. A comprehensive study is presented on the e-ciency of the proposed method. Its performance is compared with various existing techniques quantitatively and qualitatively to show its superiority in various applications.
    International Journal on Artificial Intelligence Tools. 01/2008; 17:635-658.
  • Source
    [show abstract] [hide abstract]
    ABSTRACT: Most methods for foreground region detection in videos are challenged by the presence of quasi-stationary backgrounds—flickering monitors, waving tree branches, moving water surfaces or rain. Additional difficulties are caused by camera shake or by the presence of moving objects in every image. The contribution of this paper is to propose a scene-independent and non-parametric modeling technique which covers most of the above scenarios. First, an adaptive statistical method, called adaptive kernel density estimation (AKDE), is proposed as a base-line system that addresses the scene dependence issue. After investigating its performance we introduce a novel general statistical technique, called recursive modeling (RM). The RM overcomes the weaknesses of the AKDE in modeling slow changes in the background. The performance of the RM is evaluated asymptotically and compared with the base-line system (AKDE). A wide range of quantitative and qualitative experiments is performed to compare the proposed RM with the base-line system and existing algorithms. Finally, a comparison of various background modeling systems is presented as well as a discussion on the suitability of each technique for different scenarios.
    Machine Vision and Applications 01/2009; 20:395-409. · 1.10 Impact Factor
  • Source
    [show abstract] [hide abstract]
    ABSTRACT: Detecting regions of interest in video sequences is one of the most important tasks of most high level video processing applications. In this paper a novel approach based on Support Vector Data Description (SVDD) is presented which detects foreground regions in videos with quasi-stationary backgrounds. The SVDD is a technique used in analyt- ically describing the data from a set of population samples. The training of Support Vector Machines (SVM's) in general, and SVDD in particu- lar requires a Lagrange optimization which is computationally intensive. We propose to use a genetic approach to solve the Lagrange optimiza- tion problem more e-ciently. The Genetic Algorithm (GA) starts with an initial guess and solves the optimization problem iteratively. We ex- pect to get accurate results, moreover, with less cost than the traditional Sequential Minimal Optimization (SMO) technique.
    Advances in Visual Computing, Third International Symposium, ISVC 2007, Lake Tahoe, NV, USA, November 26-28, 2007, Proceedings, Part II; 01/2007

Full-text (2 Sources)

Available from
Nov 30, 2012