A Lane Identifying Approach of the Intelligent Vehicle in Complex Condition

Abstract

Lane identification by intelligent vehicles in some common complicated conditions is a key technique. The parabola model of lane boundary is established first for fitting the curved lane boundary, and the objective function is used to estimate the fitting quality of the lane boundary is designed with the grayness character and grads character of the lane boundary in a gray image. To improve the speed of the identifying algorithm, the ant colony algorithm is applied to lane identifying, and with the guidance of the designed objective function, the positive feedback of hormone will consistently allow the search the best path within the searching region of interest as soon as possible, thus, the lane boundary is found. The method can effectively eliminate the negative effect of illumination conditions or shadows of trees, etc., and identify linear or curving lane boundaries exactly. The pretreatment to image is found and the searching region of interest is set, so the real-time identifying is greatly improved.