Vehicular Movement Tracking by Fuzzy C-means Clustering of Optical Flow Vectors

Abstract

Vehicle Tracking and Detection has many useful applications, ranging from Automated Parking Management, to a huge-scale Automated Traffic Violator Apprehension System. For vehicle tracking which this study focuses on, it proposes the use of optical flow for identifying movement of particles, as well as Fuzzy C-means clustering on optical flow output, in order to separate disconnected moving vehicles from one another. By the end of the study, the proposed algorithm made use of the previous vehicle count from the detection algorithm and was able to successfully identify the location of the different vehicles within the frame.