Velocity detection based on dark channel prior and camera fixed in the middle of road

Abstract

A monocular system fixed in the middle of road for vehicle’s velocity detection within all weather conditions based on dark channel prior in fog is presented in this paper. It is divided into four major steps. Firstly, current traffic video frame will be recognised by weather recognition system based on K-NN algorithm. Secondly, vehicles will be extracted by background difference technology and tracked by camshift algorithm. Initial points for camshift algorithm are block’s centres. Thirdly, traffic lane location will be detected, saved and updated; scene distance reference (SDR) image in homogeneous fog weather will be got based on dark channel prior. Finally, vehicle’s velocity in all weather condition will be calculated. In experiment period, velocity data for nine vehicles which stand for three classic weather conditions are listed in the form of graphic and table. Experimental results verify the effectiveness of our algorithm.