Identification and Ranking of High Pedestrian Crash Zones Using GIS

The continued improvement and application of Intelligent Transportation System (ITS) data collection technologies has generated higher requirements on the hardware and software capabilities to store, transmit and process the data. A cost -effective approach to enhancing such required capabilities is to use an ITS data compression technique. T he purpose of an ITS data compression technique is to effectively compress and reconstruct the ITS data for being easily archived and transmitted. Nowadays, the signal-processing techniques have been widely used for the general data compression purposes. T hese techniques have a great application potential to the ITS data. This paper is intended to develop a new ITS data compression and reconstruction approach based on Wavelet Transform, Discrete Cosine Transform (DCT), Quantizing and Coding techniques in the signal-processing field, by designing the appropriate feature-distilling threshold and quantizer. The objective of the compression is to minimize the data redundancy, keep the maximum amount of useful information, limit the level of distortion and improv e the data quality and application efficiency. To illustrate the effectiveness of the proposed model, the traffic flow data from the Beijing's 3 rd Ring Expressway are used for the case study. By comparing the reconstructed data with the original raw ITS da ta, it is demonstrated that the application of the proposed technique has increased the compression rate by 61.92 percent in comparison with the widely used WinZip software. On the other hand, only 5.35 percent of the data has shown a larger -than-five-percent error between the reconstructed and the original raw data. Therefore, the proposed approach has achieved a much higher compression rate with very limited level of distortion. The application of the proposed technique would contribute to the reduction o f the required storage space as well as the improvement of the transmission speed for the ITS data.