Segmentation and Tracking of Multiple Humans in Crowded Environments

AbstractSegmentation and tracking of multiple humans in crowded situations is made difficult by inter object occlusion. We propose a model-based approach to interpret the image observations by multiple partially occluded human hypotheses in a Bayesian framework. We define a joint image likelihood for multiple humans based on the appearance of the humans, the visibility of the body obtained by occlusion reasoning, and foreground/background separation. The optimal solution is obtained by using an efficient sampling method, datadriven Markov chain Monte Carlo (DDMCMC), which uses image observations for proposal probabilities. Knowledge of various aspects, including human shape, camera model, and image cues, are integrated in one theoretically sound framework. We present experimental results and quantitative evaluation, demonstrating that the resulting approach is effective for very challenging data. Existing system: The existing system detects appearance or shape-based patterns of humans directly. Those in methods learn human detectors from localshape features and build contour templates for pedestrians. These learning-based methods need a large number of training samples and

may be sensitive to imaging viewpoint variations as they learn 2D patterns. Besides motion and shape, face and skin color are also useful cues for human detection, but environments where these cues could be utilized are limited, usually to indoor scenes where illumination is controlled and the objects are imaged with high resolution, Without a specific model of objects, tracking methods are limited to blob tracking. Proposed system: The goal of this work is to develop a method to detect and track humans in the presence of persistent and temporarily heavy occlusion. We do not require that humans be isolated, that is, unconcluded, when they first enter the scene. However, in order to see a person, we require that at least the head-shoulder region must be visible. We assume a stationary camera so that motion can be detected by comparison with a background model. We do not require the foreground detection to be perfect, e.g., the foreground blobs may be fragmented, but we assume that there are no significant false alarms due to shadows, reflections, or other reasons. We also assume that the camera model is known and that people walk on a known ground plane.

Block diagram:

Tool used: 1. Matlab 2010b