Human tracking system

Abstract

This paper presents a comprehensive framework for tracking coarse human model from sequences of synchronized monocular grayscale images in single or multiple camera coordinates. It demonstrates the feasibility of an end-to-end person tracking system using a uniqu e combination of motion analysis from sequences of synchronized monocular grayscale images in differe nt camera Coordinates and other existing techniques in motion detection, segmentation, and patter recognition. This human tracking is an important task in many vi sion applications. The main steps in video analysis are two: detection of interesting moving objects and tr acking of such objects from frame to frame. In a similar vein, mos t tracking algorithms use pre-specified methods for preprocessing. There are several objects tracking algorithms i.e . Meanshift, Camshift, Kalman filter with different preprocessing methods. The system starts with tracking from a single camera view. When the system predicts that the active camera will no longer have a good view of th e subject of interest, tracking can be switched to another camera which provides a better view and requires the least switching to continue tracking. The nonrigidity of the human body is addressed by matching points of the middle line of the human image spatially and temporally, u sing Bayesian Classification schemes. Multivariate normal distributions are employed to model class-conditi onal densities of the features for tracking, such as loc ation, intensity, and geometric features.