基于随机Hough变换的人头检测

传统的人头检测方法多为基于人脸和头发的检测，误差较大。为此，提出一种基于随机Hough变换(RHT)的人头检测方法。根据人头部轮廓近圆的特征，采用Canny算子提取图像边缘，得到目标轮廓。利用RHT算法对独立的曲线进行圆检测，并对人头进行标识。实验结果表明，与现有方法相比，该方法的识别率较高、速度较快、适用范围较广。

Head Detection Based on Randomized Hough Transform

The principles of traditional head detection methods are mostly based on the detection of the face and the hairs,these methods have error.Aiming at this problem,this paper presents a head detection method based on Randomized Hough Transform（RHT） focusing on the feature that the contour of human＇s head is similar as a circle.Object＇s contour is obtained by edge detection based on canny operator.The circles are detected by the method of RHT in every independent curve,and the heads are marked.Experimental results demonstrate that this method has higher accuracy,lower time cost and greater scope of application in contrast with the existing methods of head detection.