Abstract

One of the most emergent research is suspicious behaviour monitoring in video surveillance. In recent past, crime detection is powerful topic to identify the abnormal events or crime events. This work focused on the suspicious behaviour analysis which helps to detect the crime events in terms of gait parameter. This work describes the following tasks. First, tracking the pedestrians from video data using MM track algorithm i.e. (calibration process). Second, extracting the gait parameters based on proposed modules: 1) spatial coordinate module contains the speed profiles which helps to measure the suspicious behaviour of pedestrian. 2) Fixed coordinate system module, it also measures the suspicious behaviour in different way based on the list of components and axis of the pedestrians. This step performs the major role in measure the suspicious behaviour among the pedestrians’ movement for crime detection. Third, measure the suspicious behaviour in terms of walk ratio, Acceleration Auto Correlation (AAC) and gravity, dynamic, horizontal, vertical components of pedestrians as well this step ? value performs the validation role which is based on the reference range to validate the Walk Ratio value. The video helps to monitor the pedestrian’s movement. This work is compared to the different pedestrian’s detection technique such as DPM (Deformable Part Model) and Real Boost method foe efficiency in terms of true positive rate and pedestrian gait speed detection time parameters. Proposed work attains best result in both parameters.