A traffic incident detection method based on wavelet Mallat algorithm

Abstract

For aim applied to develop intelligent transportation system and the characteristic of traffic flow breakdown, a traffic incident detection method based on fast Mallat algorithm of wavelet analysis is presented. Utilizing the association between the wavelet coefficients and traffic flow, the condition of traffic flow can be extracted directly from the approximate coefficients and detail coefficients of wavelet decomposition from traffic flow parameters. Using data obtained from the simulation under the condition of incident and non-incident, parameters of the algorithm are calibrated and an off-line test is made. According to results of the test compared with California algorithm, low-pass algorithm and MLF algorithm, the algorithm performs better than the other algorithms in traffic incident detection.