Human Detection/Tracking System for Video Surveillance With Noise Removal

Abstract

Object detection and tracking have been extensively used in many applications including security and surveillance. This chapter addresses the problem of human detection and tracking in surveillance videos with noise. The system proposed deals with video processing utilizing Kalman filtering to enhance the process in the presence of challenging weather conditions. To support the target idea, some experiments are introduced to measure the deviation between both the noiseless and the noisy videos and to study the effect of each filter within each noise type of video disturbance. The most efficient masking system is used to enhance the outcome in the presence of video degradations. The proposed tracking and detection technique achieves achieve an average correct assigned tracks accuracy of about 95% for noiseless video. This is due to the challenging video degradation depending on the type and level of noise. In case of noisy videos by applying the correct filter mask, the accuracy of tracking in the presence of a median filter comes to around 90%.