Vehical Tracking and Speed Estimation of Moving Vehicles for Traffic Surveillance Applications

Now-a-days traffic in two-tier cities has increased lot, so surveillance system is needed to monitor the traffic and avoid unwanted delays and accidents. Vehicles speed estimation can be done by analyzing the video captured. The conventional or today’s method used to control or measure the speed of the vehicles is through using RADAR device and other models like constant g-factor, sequence method and motion median. The limitations of this device or models are accuracy is bit less and costly and does not capture long distance images. Different techniques have been used for speed estimation; here in this paper we propose a mixture of Gaussian technique to estimate the speed. In speed estimation we need to detect vehicle from the captured video and analyze the video considering the centroid at each instance of a frame and distance will be calculated. The proposed method consists of mainly four phases background subtraction using Mixture of Gaussian, feature extraction using Shi-Tomasi technique, vehicle tracking, distance and speed calculation. The effectiveness of using Gaussian mixture model was evaluated with a sample video data set and the actual speed and the determined speed comparison is shown in the below results. The vehicle speed is estimated using distance travelled by the moving vehicle over number of frames and frame rate.