Improved Algorithm for Object Tracking in Video Camera Network

Objectives: To develop a novel approach for tracking the object in multiple camera environments. The object tracking in situation of suspect person is the challenging task for a researcher for a long time. The situation get worst if the object hided beside any other object, or if the object moves from the range of camera span Methods/Statistical Analysis: The object is tracked using Optical Flow Estimation techniques in which magnitude along the gradient direction is observed. The threshold magnitude comparison of the images detects the desired target. Findings: Detection parameters in the our proposed methods such as SIMILARITY, F1, PRECISION, RECALL are better in comparison with the existing methods in existing methods viz RADCT, MSDE, SDE and SSD. The proposed algorithm can easily detect object even if it is either hided or it moves to the frame of camera 2, or any other camera installed within the same camera network. Application/ Improvements: The proposed methods tracked and detects the suspected object target for the application of safety and security system more accurately and better ease of implementation than other existing motion detection system such as RADCT, MSDE, SDE and SSD.