Development of a mobile vehicle classification system

Abstract

The requirement of non-invasive counting and classifying devices has grown significantly in the last few years due to contributing factors from Occupational Health and Safety Standards developed by state road authorities. There is significant risk associated with departmental personnel working in proximity to traffic. The risk is elevated further for these personnel working on the road whilst installing the devices in high speed and high volume traffic environments. The focus of this project was to develop a technology capable of classifying vehicles into the Austroads 94 12 Bin Classification Standard. The restrictions for the system were such that no devices or apparatus are placed on the carriageway, the detection system would not affect motorists in anyway and the system would be portable and suitable for multilane high-speed roads. The system chosen to develop was based on the use of 2 infrared lasers. The project deals with noise in the laser data via comparison of denoising techniques such as Low Pass Filtering, Frequency Domain Thresholding and Discrete Wavelet Transform. It was found in the project that Wavelet Denoising offers the best noise reduction for the signal and was far superior at retaining the signal properties. The accuracy of the developed system was found to be low compared with that of commercially available systems. However, the discrepancy is attributed to sensor synchronization, a problem able to be solved by the introduction of a designated processor.