Research Of Vehicle Recognition Based On Audio Feature Analysis

The implement of intelligent transportation systems is critical for the development of intelligent transport systems, the key lies in the detection and identification of the vehicle. While the mainstream of the detection methods currently used, for various reasons, is still difficult to meet the requirements of a large number of settings along the road, and therefore, the paper with the audio signal generated by the vehicle is stationary or when, based mainly on the audio signal characteristic of the vehicle is analyzed in based on the theory proposed vehicle audio signal for identifying the vehicle on the basis of this program, and the initial identification. Research work are as follows:(1) Based on the idea of system development, the overall design of the system is described, detailed description of the software architecture of the system as a whole, including the audio feature extraction module, audio features based on the vehicle identification module design ideas and describes the system configuration database.(2) The audio feature extraction module is designed indetail. First, in order to improve the algorithm for different SNR noisy speech processing capability, the advantages of Wiener filtering is combined with adaptive filtering, improved spectral subtraction; secondly n order to ensure the system recognition performance, the Mel-cepstrum investigates are selected with a first-order differential Mel-cepstrum as feature parameters, to a certain extent, improve system robustness.(3) Recognition model is designed, in-depth study of probability models Gaussian mixture model approach, GMM can not only take advantage of the dynamic information of the speech signal timing, the impact of endpoint detection accuracy of its recognition performance is also very small, designed and implemented based on the GMM vehicle identification methods.(4) Software functionality and retrieval algorithms designed in the previous chapters were carried out pilot testing and software testing, simulation experiments, and draw the relevant conclusions.