基于边缘背景差法和Hough变换的公交乘客头部检测方法研究

Bus Passengers' Head Detecting Based on Edge-Based Background Subtraction and Hough Transform

为提高公交乘客检测的准确率,介绍了一种改进的背景差法——将当前图像和背景图像提取边缘再做背景差。该方法受光照条件变化和干扰物的影响较小,能够有效提取目标边缘;在此基础上通过Hough变换获得乘客头部的圆形轮廓,然后从视知觉角度分析并除去检测到的虚假轮廓,从而实现乘客头部的识别。实验结果表明,该方法简单,易于操作,识别率高,为公交乘客流量统计奠定了良好的基础。

In order to improve the detecting rate of bus passengers,the background subtraction method is proposed,which subtracts the edge of the background from the edge of the present image.This method is influenced little by the change of light conditions and could extract the edge of the object efficiently.Hough transform is used to get the circle figure of the passengers,and then false figures are deleted through analysis on perceptual point,in order to detect the head of the passengers.The results of tests show that this method is simple,easy to operate,and the detecting rate is high,which affords a good foundation for the automatic passenger counting.