Research On Key Technologies Of Parking Space Detection Based On Machine Vision

With the development of intelligent vehicle technology,and the demand for intelligent parking system,automobile manufacturers and research institutions carry out the research and development of automatic parking system.Automatic parking technology can assist people to effectively solve the parking problem,and parking space detection is the first problem to be solved in automatic parking system.This paper research on key technologies of parking space detection based on machine vision,the specific research contents and methods as follows:The primary goal of parking space detection is parking lines detection,road extraction is to reduce the parking lines detection interference caused by vehicle and the background environment,then detect the edge through Canny,carry on Hough transformation,using the least squares fitting the segment lines which locate on the same line,extract the approximate parallel lines according to the double lines characteristics of parking lines,using the extracted double lines tectonic T sign which can represent the parking line features,according to the position of constructed T sign locate to the original image area,extract the interest region of the original image,through the color similarity judgment method checking the interest region whether is T sign,thus confirming the extracted approximate parallel lines are parking lines.This paper studies two cases parking lines including the solid and dashed lines,the experiments show that,this paper's parking lines detection method takes on well accuracy.The research object of this paper are the vertical and inclined parking spaces in the parking lots,parking type determines the method of path planning in automatic parking system,therefore,the automatic classification of parking space is particularly important,according to T sign angle of the parking lines detected in the inverse perspective image,distinguish the parking is vertical or inclined.This paper's parking space detection has also includes the effectiveness determination of parking spaces,parking lines top line in the image does not have double lines characteristics,connected parking lines four sides together to form a closed parking line,and fitting it to the binary image of the road extraction,according the occupancy of the road surface pixels of closed parking lines to judge the parking spaces is empty or not empty,and makes the identification.Based on the above algorithm,parking space detection based on machine vision system is developed using Microsoft Visual C++6.0 and OpenCV library,the experiments show that,The method deal the collected parking spaces videos and images with good result.