Application of Discrete Wavelet Transform to Traffic Measurement Using Road Environmental Sound from Microphone on the Vehicle

摘要

One of the enabling technologies for ITS is traffic How measurement. Since current traffic flow measurement relies on local sensors, the development of image sensors that can treat information over wide areas has been undertaken by a number of organizations. Similarly, development of acoustic sensors is desirable, although no practical products have been realized. Few research examples on acoustic sensors exist. Recently, research on probe vehicles, in which ambient information is collected by on-board sensors, has been carried out extensively. Improvement of sensing technologies using mobile sensors is highly desirable. The present paper reports the results of traffic flow measurement that was conducted by counting the number of oncoming vehicles with on-board mobile acoustic sensors. By means of the discrete wavelet transform, the road traffic environment noise is divided into five levels. The component of each level is compared with the noise without oncoming vehicles. The variations are represented as the vertices of a pentagon. The oncoming traffic situation is judged by the change in the area of the pentagon. Experiments show that it is possible to count the number of oncoming vehicles with a recognition rate of 75.5%. Future prospects are discussed.