智能车行人识别的灰度形态学边缘检测研究

Grayscale Morphology Edge Detection for Pedestrian Recognization of Intelligent Vehicle

行人轮廓特征是行人识别的重要特征,为了获取更好的行人轮廓,边缘检测方法是一个非常重要的选择。首先采用Roberts算子、Prewitt算子、Soberl算子、LOG和Canny算子对行人图像进行边缘检测,得出能获得较好行人轮廓的边缘检测方法;然后用此边缘检测方法结合灰度形态学进行图像处理,最终获得具有最优行人轮廓的形态学边缘检测方法。试验结果显示,此方法能很好地消除行人轮廓的干扰因子。

Pedestrian profile is an important feature of pedestrian recognition. Edge detection is a very important choice to get a better pedestrian profile. First, using Roberts, Prewitt, Soberl, LOG and Canny to detect the edge of the pedestrian and obtaining the method of edge detection for getting a better pedestrian profile. Then combining with grayscale morphological for image processing. Finally getting grayscale morphology edge detection method that could get optimal pedestrian profile. The results show that the method can be used to eliminate the interference factors of pedestrian profile.