Application and Analysis of Time Domain Cross Correlation for Traffic Flow Speed Measurement

Abstract

Radars are common means to measure vehicle speed. They are fast and accurate, however measure only instantaneous velocity not speed of the traffic flow. Therefore they are not suitable to help determine the busy degree of traffic. A novel method of traffic flow speed measurement using video based on time domain cross correlation is introduced. It provides 'flow speed', a much better parameter determining whether there is a traffic jam or how fast one can cover that road under the observed situation. With regard to the 'False-Peak' problem, effect of the sequence length on correlation result is presented, together with formulas used to calculate proper sequence length both accurately and approximately. Test results demonstrate the signification of sequence length on improving cross correlation result and good accuracy suitable for estimation of road busy degree.