Moving Object Detection and Tracking Using Hybrid Model

Multiple moving object detection in videos is very important in many video processing applications like video surveillance, monitoring traffic for rash driving control, detections of pedestrians etc. Thus detection must be performed accurately and robustly to minimize the false alarms and missing evidences. Background modelling and foreground extraction are two major processes to achieve this goal. Many traditional background modelling methods use either colour information or texture information. But colour is sensitive to light variations and texture information cannot be utilized to separate smooth foreground from smooth background in many cases. To achieve good performance in terms of high foreground detection accuracy and low computational cost is also challenging. A new hybrid model with integration framework of texture and colour information for background modelling is proposed in this project. This framework is able to combine the advantages of both colour and texture methods, and at the same time it cancels out their disadvantages as well. Moreover, we propose a block based method to accelerate the background modelling. The background and foreground models are updated by first in first out strategy to maintain the most recent observed background and foreground instances. Along with this it is necessary to track the detected objects in real time, to enable corrective actions. We are using active contour model based tracking. An extensive experiment on various challenging videos and comparison of various parameters like Precision, Recall, F-measure and Processing time which proves the effectiveness of the proposed method over existing ones.