Sobel-Hough检测方法在机器视觉倒车辅助系统的应用研究

Detection of obstacles' edges based on sobel-hough algorithm by parking assist system

倒车辅助系统,又称泊车辅助系统或可视倒车雷达,能够给驾驶员倒车泊车操作带来很大的方便。对主流倒车辅助系统的研究后,引进机器视觉,为精确检测柱子类或台阶类障碍物提供了新方法。倒车操作时,使用CCD内核的视觉传感器获取车后的图像信息,将彩色图象进行中值滤波后转变为灰度图像,采用Sobel算子锐化突出柱子或台阶的边缘,再采用自适应阈值法分割成二值图象,接着用Hough变换方法检测出边缘线段,最后对障碍物边缘进行加粗加色标记。实验表明,采用Sobel-Hough算法能准确地将柱子和台阶边缘识别出来,并具有较强的抗干扰能力。

Parking Assist System can bring the great convenience of backing and parking the vehicle to drivers.a new method how to detect and get the edge of obstacles is introduced.The CCD sensor which is behind the boot of car gets the picture of obstacles which usually are poles and steps.After processing the picture by median filter,the color image translates into the black-and-white image,and the edge detection algorithm Sobel and adaptive global thresholding are applied to get the coarse edges.Then the Hough transforms algorithm is applied to locate the most important edges.At last,the most important edges will get wider and be colored.The original picture with the colored edges will be shown in the screen of the car.