Frame feature tracking for speed estimation

Frame feature tracking, relative pixel distance, relative pixel direction measurement and conversion of pixel distance to real world distance are the techniques used to estimate the speed of the vehicle. The balance between the performance and accuracy are the critical tasks in many existing speed detection techniques. In this paper, a novel speed estimation technique using feature tracking is presented. The main goal of this research is to increase the accuracy of the speed estimation while having an acceptable performance level. The proposed technique use feature tracking of subsequent image frames and estimate the speed using relative distance of the features. Experimental results with several traffic situations revealed that new technique has the desired accuracy with a satisfactory level of performance.