Lane Line Detection Based On Digital Image Processing

With the rapid development of national economy, the number of vehicles isincreasing rapidly. However at the same time, the contradiction between people,vehicles, roads are also increasing. In order to solve this contradiction, the IntelligentTransportation Systems (ITS) concept came into being. This article selects one of themost critical problem in ITS what is the lane line detection as the research object.This paper structured the road lane line detection in image facing the aboveproblems, put forward two different detection algorithm based on feature.1. Edge detection-based and minimum variance method with in class werecombined for the lane line detection. This method can detect the straight lane quicklyand effectively under different conditions. The main idea is minimizing the intra-classvariance of the gradient amplitude and gradient direction angle of each pixel in astraight lane line. Firstly, The algorithm uses the edge detection method to calculatethe gradient amplitude and gradient direction angle of edge pixels for achieving theedge of the road image; and then labeling connected domain on edge image afterprocessing binarization; When extracting straight lane line, using the minimumintra-class variance method to calculate the minimum variance of gradient directionangle on boundary points of each connected domain, finally setting threshold to isolatethe linear lane line, and achieving the goal of the lane line detection.2. Based on the integration of the improved k-means clustering algorithm and thecustom edge detection method to detect the lane line. This method can effectivelyovercome the influence of uneven illumination, rain and snow, etc. At first smoothhistogram method is used to determine the number K of k-means cluster, making thelane can be accurately clustered into the same category, then performing image fusionwith the image edge extracted by using a custom operator and cluster image, makingfull use of the characteristics of the lane line edge, obtaining the lane line area. Through classical linear extraction method Hough transform can locate the lane lineaccurately and complete the lane line detection.Finally, two detection algorithms by a large number of simulation experimentsconfirmed the effectiveness of the two algorithms.