Determination of Road Traffic Flow Based on 3D Daubechies Wavelet Transform of an Image Sequence

摘要

Daubechies wavelet transform is proposed to represent the contents of the image sequence. In order to account for temporal changes of the contents a 3D transform is used. 3D DWT, Daubechies based, is chosen for calculations to determine the coefficients. A method of mapping road traffic flow is developed. The method uses a linear function of the wavelet coefficients for describing the changes in detection fields, which is in turn is converted to traffic flow. The parameters of the linear function are determined by minimizing the MSE of fitting this function to corresponding traffic flow values. The method is validated using a set of video sequences.