Research On Lane Line Detection Algorithm Based On Digital Image Processing

The lane line detection technology is widely used and can be used for intelligent traffic control and safe assisted driving.The unmanned technology that emerged in recent years is also based on the new technology derived from the detection technology of lane lines,so the study of problem of lane detection has a strong social utility.At present,most researchers are researching and improving various algorithms to enhance the effect of lane detection,but there are still two problems that are not solved well: the first problem is that the proposed algorithm is not real-time enough to meet the requirement of real-time line detection system;the second problem is that the accuracy of the detection is not high enough to meet the accuracy requirements of the lane detection system.This paper mainly focuses on lane line identification and detection,and focuses on solving the problem that the real-time of lane line detection algorithm is not strong and the accuracy is not high.Firstly,the road image is captured and acquired,and then the captured image is preprocessed.The preprocessed process includes segmentation of the region of interest,image graying,image filtering,and edge enhancement.Specifically,(I)combined with the actual shooting environment,divide the captured image into five equal parts,and take the following 3/5 part as the region of interest.(II)In order to remove the irrelevant information of the color components in the image and improve the running efficiency of the subsequent algorithm,the image is grayscaled by the weighted average method.(III)In order to remove the extraneous noise introduced during the shooting process,and improve the visibility of the image and the captured image is filtered.The median filtering is used as the filtering method.(IV)In order tohighlight the region of interest and reduce the information of the region of no interest,the image is edge-enhanced,that is,the image is edge-enhanced by the histogram equalization method..Next,in order to facilitate subsequent lane line detection,edge detection is performed.The four edge detection algorithms include Canny edge detector,Kirsch edge detector,Prewitt edge detector and Roberts edge detector are introduced in this paper.Each algorithm is simulated.After comparison,Kirsch edge detector has the best performance,but its shortcomings are also obvious,that is,the amount of calculation is too large and the operation time is too long.The detection algorithm is improved by using the correlation between the upper and lower elements in the eight template matrices used in the Kirsch edge detector.The computational complexity is greatly reduced,and the real-time detection is improved a lot.Finally,the algorithm of Hough transform detection line is introduced and improved,and the flow and steps of detection of the algorithm are introduced.Hough transform is an important choice for detecting lane lines at present,but its computational complexity is difficult to meet the real-time requirements of the system.In order to improve the real-time performance of the Hough transform algorithm,the Hough transform is improved by the property of the additivity of the Hough transform.The image to be detected is divided into several sub-image blocks.The Hough transform problemwhich originally requires all pixels on the entire image becomes a Hough transform problem in which a partial pixel of a sub-image block.,the calculation amount is greatly reduced,and the real-time performance is greatly improved.After experimental simulation,the results show that the improved Hough transform reduces the running time by 70% compared with the traditional Hough transform,and the error rates of the two most important parameters ? and ? in the Hough transform are also acceptable.It can meet the requirements of real-time and accuracy of the lane line detection system.Compared with other edge detection operators,the improved Hough transform algorithm combined with the improved Kirsch detector reduces the relative error rate of the angle parameter ? by 2.38% and the relative error rate of the distance parameter ? by 2.34%.Compared with other classical algorithms,the improved Hough transform algorithm has relatively obvious advantages in both real-time and accuracy.The operation time is between 1/3 and 1/2 of other algorithms,the relative error rate of the angle parameter ? is reduced by 2.78%,and the relative error rate of the distance parameter ? is reduced by 1.02%.The main contribution of this paper is to improve the Hough transform algorithmby using the additivity of the Hough transform,and to improve the Kirsch edge detector.Regardless of real-time or accuracy,the improved algorithm has a relatively obvious advantage,which can meet the real-time and accuracy requirements of the lane line detection system.