【摘要】 人体定位是指在图片中确定行人位置的过程。为了提高人体定位的正确率和速度,提出了一种结合小波变换和不变矩匹配的定位方法。首先对图像进行小波变换,提取出图像小波系数;再根据行人模板的系数特征提取出不变矩模板;然后在小波域上通过不变矩匹配实现人体定位。该方法减少了待检区域,降低了计算代价。实验结果表明,该算法识别率达78.6%,测试结果比Oren等人提出的匹配方法提高了26.7%。更多还原

【Abstract】 Pedestrian detection is a processing that locates the person in the image.In order to enhance the efficiency of detection,this paper presented a new method using the wavelet template compound with invariant moments matching.First the coefficients of image could be got after the wavelet transform.Then the invariant moments templates were got according to the coefficient of pedestrian.At last using the invariant moments matching we located the position of person in the wavelet image.This new method reduced the area of detection so that the cost of computation had been cut down.Experiment results showed that this algorithm achieved a detection rate of 78.6%.There was a increase of 26.7% for detection compared with wavelet coefficient matching.更多还原