基于环视图的自动泊车停车位检测方法  

Automatic Parking Space Detection Method based on Ring View

作　　者：范梦阳 Fan Mengyang

机构地区：[1]重庆电子工程职业学院,重庆市401331

出　　处：《时代汽车》2021年第11期189-190,共2页Auto Time

基　　金：基于多传感器融合的车辆自动泊车停车位检测方法研究;编号:XJZK202002。

摘　　要：“停车难”这一问题变得日益严峻,自动泊车系统的研究也就显得至关重要,停车位检测是自动泊车系统的重要组成部分。环视图通过逆透视变换消除了透视畸变,恢复了库位线的平行、垂直、间距等几何信息,可以直接利用图像中的库位线满足的几何约束约束特征来实现停车位检测。利用安装在车辆四周的四个鱼眼摄像头采集图像,经过畸变矫正、逆透视变换、重合区域拼接可以得到车辆的环视图。摄像头在采集图像的过程中必然会引入各种噪声,首先需要对环视图通过均值滤波、中值滤波、高斯滤波等方式进行降噪处理。之后,进行灰度化,以减少后续图像处理的计算量。由于库位线与地面的灰度信息不一致,灰度化后的图像仍然保留了库位线的灰度特征,仍然可以利用该灰度差异区分出停车位线和地面。在对环视图进行滤波降噪、灰度化等图像预处理操作之后,通过边缘检测得到库位线的边缘。之后利用霍夫变换提取出边缘图像中的直线。根据《中华人民共和国国家标准/道路交通标志和标线(第3部分:道路交通标线)》(GB5768.3-2009)标准中对停车位标线宽度、尺寸和设置方式的规定,提取划线停车位应满足的几何约束特征条件,可以识别出停车位。在此基础上,融合车辆定位模块得到的车辆位姿信息对单帧图像中检测到的停车位在图像序列中进行跟踪,进一步提高停车位检测正确率和检测精度。The problem of"difficult parking"is becoming more and more serious,and the research of automatic parking system is also important.Parking space detection is an important part of automatic parking system.The ring view eliminates the perspective distortion through the inverse perspective transformation and restores the geometric information such as the parallel,vertical,and spacing of the location line.It can directly use the geometric constraints that the location line in the image meets to realize the parking space detection.The four fisheye cameras installed around the vehicle are used to collect images,and the ring view of the vehicle can be obtained after distortion correction,inverse perspective transformation,and overlapping area stitching.The camera will inevitably introduce various noises in the process of collecting images.First,the ring view needs to be denoised by means of mean filtering,median filtering,and Gaussian filtering.After that,grayscale is performed to reduce the amount of calculation for subsequent image processing.Since the gray information of the location line is inconsistent with the ground,the gray-scaled image still retains the gray features of the location line,and the gray level difference can still be used to distinguish the parking space line and the ground.After image preprocessing operations such as filtering and denoising and graying are performed on the ring view,the edge of the location line is obtained through edge detection.Afterwards,the Hough transform is used to extract the straight lines in the edge image.According to the"National Standards of the People’s Republic of China/Road Traffic Signs and Markings(Part 3:Road Traffic Markings)"(GB5768.3-2009),the rules for the width,size and setting method of parking space markings are extracted.The geometric constraint characteristic condition that the parking space of the line should meet,and the parking space can be identified.On this basis,the article fuses the vehicle pose information obtained by the vehicle positioning mod

关 键 词：自动泊车 停车位检测 摄像头 环视图 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]