基于车载辅助驾驶平台的快速行人检测技术及其应用-智能科学与技术专业论文.docxVIP

优秀毕业论文 精品参考文献资料 AbstractPedestrian Abstract Pedestrian detection is a basic and important research task in computer vision．It is widely used in many fields，such as automobile assistant driving，video surveillance， robot and SO on．Pedestrian detection system is very important in automobile driving． However,due to the complexity of the vehicle pedestrian detection,the situation is changeable，and the real-time requirement is high，the pedestrian detection is faced with many challenges．How to obtain good detection effect in real-time condition is the core of the research on vehicle pedestrian detection,and is also the key point of this paper． In this paper，the corresponding solutions are proposed based on the two major problems of vehicle pedestrian detection accuracy and real-time performance．The main work and innovation of this paper are as follows： (1)In this paper，the vehicle pedestrian detection is performed by using the aggregated channel feature and the AdaBoost classification method．The feature representation of the aggregated channel is strong，its computational complexity is low,and the AdaBoost method is fast，and it is suitable for real—time target detection． In this paper，this method is first applied to vehicle pedestrian detection． (2)In this paper,the detection method is improved according to the particularity of the vehicle pedestrian detection task．The detection Was performed by using multiple model scale and multi aggregation scale．This method gives consideration to bom the accuracy and the real-time performance，and it is almost 3 times faster than the original method in the same precision． (3)In view of the problem that the deformation caused by the deformation of pedestrian in the moving process，this paper introduces the pedestrian tracking． Through the method of tracking and detection,the detection accuracy is improved effectively．Due to the high real—time tracking processing，the detection frequency Can be reduced，and the detection time Can be further