基于混沌粒子群的车道线检测算法

Lane detection algorithm based on chaotic particle swarm

为解决在恶劣天气、光照条件变化和路面信息干扰等情况下的车道线检测识别率低的问题,提出一种采用核Fisher线性判别分析灰度变换以及混沌粒子群的车道线检测算法。将高维的RGB彩色图像通过最佳鉴别投影向量投影到低维的子空间,通过混沌粒子群算法,根据车道线特征,遍历粒子取值范围内的空间,寻找适应度函数最大的解,根据最优解获得的直线参数在图像上拟合车道线。实验结果表明,所提算法能够实现各种道路情况下的车道线检测功能,验证了该算法具有良好的鲁棒性。

To improve the lane detection accuracy in the case of bad weather,the changing in lighting conditions,the road information interference and so on,a lane detection algorithm combining the kernel Fisher discriminant analysis gray scale transformation and chaotic particle swarm was proposed.The high-dimensional RGB color image was projected to the low-dimensional subspace using the optimal discriminant projection vector.The chaotic particle swarm algorithm was used to traverse the space within the range of the particle value according to the lane line feature,and the largest fitness function was found.The lane marking was fitted on the image according to the line parameter obtained according to the optimal solution.Experimental results demonstrate the lane marking detection can be identified using the proposed method under various road conditions with good robustness.