Vehicle Pedestrian Detection System Based On DM6437 Optimization And Implementation

Vehicle infrared pedestrian detection system based on DM6437, to extract and determine in infrared image pedestrian information by means of image processing and pattern recognition technology, a danger to pedestrians to remind the driver in front of the possible. Far infrared camera is based on thermal radiation imaging, is not influenced by the external light source, particularly suitable for use in the heavy fog weather at night or under. The work of this paper is based on preliminary laboratory on the engineering, mainly solves the problem of pedestrian detection feature extraction due to simple and leads to high false alarm rate is not high, the realtime problem of optimization, and the algorithm of pedestrian detection due to the insufficient difference. The main work of this paper is as follows:1) In-depth understanding of the DM6437 chip features, multi-tasking and synchronization mechanism for understanding DSP/BIOS. On this basis, puts forward and realizes the input infrared image is parity frame processing strategy and frame processing strategy, based on the system detection rate, improve processing speed on the infrared image system, which improves the image output frame rate.2) Learning and using the TI VLIB library and IQmath library and the DSP platform features, to optimize the performance of the system, improve the real-time performance of the system.3) In the realization of pedestrian detection in the main program, design and realize the target of pedestrians and background ROI effective cascade classifier algorithm, including the detection of target aspect ratio, the scene information filtering, adaptive detection, road detection head characteristics of +SVM classification based on HOG. Also on the pedestrian ROI distance were predicted, the detection region division system in order to focus on regional and non-regional focus, to take on the pedestrian different sensitivity algorithm in each region.4) Communication by using UART to design and implementation of PC terminal and DSP terminal, to realize the function of man-machine interaction.5) On different scenarios for system test scale of system. Testing is mainly from the detection rate, false alarm rate, the three aspects of real-time. The test results show that the system can basically meet the requirement of practical application, could push the market.