【摘要】 为能在单目图像中快速准确地检测出行人,提出一种基于ROI提取的行人检测算法。根据待检测图像的边缘信息提取所有可能含有行人的区域,用区域的中心点代表该区域;基于这些中心点,根据行人在图像中固有的特点,确定行人ROI区域;对该ROI区域进行行人的检测。在INRIA测试集上的测试结果表明,该算法在误检率较低、检测速度较高的情况下,检测率达到95%以上,可以用于实时行人检测系统。更多还原

【Abstract】 To detect pedestrian in monocular images quickly and accurately,apedestrian detection algorithm based on ROI extraction was proposed.Based on the edge information of image,the area which might contain pedestrian was found and the middle point was used to represent the area.According to the points and the characteristics of pedestrian in the image,the ROI was determined.Pedestrian was detected only in these ROI.Experimental results on the test set of INRIA show that,the algorithm can improve true positive rate to 95% and has higher detection speed,lower false positive rate than other algorithms and can be used in real-time system.更多还原