BACKGROUND MODELING AND SUBTRACTION FOR PROACTIVE SURVEILLANCE OF MASS TRANSPORT SYSTEM

Abstract

The problem of visual inspection of mass transport systems (e.g. airports, railway stations, roads etc.) has received growing attention in recent times. Traditionally, the most important tasks of surveillance & monitoring safety are based on human visual observation; however, an autonomous system able to detect anomalous or dangerous situations can help a human operator, even if it cannot completely replace its presence. The purpose of intelligent surveillance systems is to automatically perform surveillance tasks by applying cameras in the place of human eyes. Recently, with the development of video hardware such as digital cameras, video capture card etc., the video surveillance system is becoming more widely applied and is attracting more researchers to develop fast and robust algorithms. All Automated surveillance systems require some mechanism to detect interesting objects/ abnormal activities in the field of view of the sensor. The first image processing step in this regard is the background modeling of the scene. In this step, foreground moving objects are segmented out from the background. In this dissertation work, we present an algorithm for foreground-background segmentation which is based on codebook-construction method. In this method, sample background values at each pixel are quantized into codebooks which represent a compressed form of background model for a long image sequence. This allows us to capture structural background variation due to periodic-like motion over a long period of time under limited memory. The codebook representation is efficient in memory and speed compared with other background modeling techniques. The method can handle scenes containing moving backgrounds or illumination variations, and it achieves robust detection for different types of videos. Here, we present an algorithm for foreground-background segmentation by codebook using automatic threshold detection. The proposed algorithm automatically calculates the threshold value for each codeword separately and subsequently uses it in the process of background subtraction via code-book generation. Experimental results show that the proposed automatic threshold detection codebook-based background subtraction algorithm is very much suitable for the applications related to automatic video surveillance.