A Wavelet-based Algorithm for Vehicle Flow Information Extraction

Abstract

This paper proposed an improved algorithm applied in video intelligent traffic control system for vehicle detection. The accuracy of original algorithm, which is based on the comparision of contrast and luminance distortion of present image with background, reduces greatly under bad weather because of false detection caused by noises in captured images.In this paper we chose Daubechies wavelet as mother wavelet to add a 2-dimension wavelet process before the algorithm, just after the imageis captured, to de-noise each captured image. We used FPGA-based equipments to test the algorithm, and the experiment proved higher performance of improved algorithm, especially under bad weather.

http://dx.doi.org/10.11591/telkomnika.v11i1.1916