Abstract: Genetic algorithm is an efficient parallel global search method with good robustness, parallelism and adaptability, which is very suitable for the optimization of large-scale search space. In order to improve the accuracy of lane recognition, and to better protect the details of the image while effectively eliminating noise, a threshold segmentation based on improved genetic algorithm is proposed for the problem of poor adaptiveness and weak anti-noise ability of traditional image processing threshold segmentation. method. The method uses binary coding and determines the first generation population, and uses the maximum inter-class variance calculation formula as the fitness function, and uses the variance between classes as its fitness value. The threshold of image segmentation is calculated by genetic algorithm, and then the lane line is extracted. Features are curve fitted. The results show that the improved genetic algorithm for image processing can improve the anti-noise ability and has better threshold segmentation adaptability, which can effectively segment the image threshold and improve the accuracy of lane recognition. Keywords: lane line recognition; threshold segmentation; genetic algorithm; fitness; maximum inter-class variance

CICTP