LANE RECOGNITION METHOD

摘要

According to the present invention, a lane recognition method comprises: a step of converting a photographed image provided by a camera into a binary edge image; a step of processing the edge image to obtain a vanishing point and a horizon position which is a line crossing the vanishing point; a step of processing the edge image to obtain a hood position which is a line where a road and the hood of a vehicle meet; a step of using the slopes of lines and the brightness of pixels of the lines positioned between the horizon position and the hood position in the edge image to select one left lane line and one right lane line; a step of positioning the one left lane line and the one right lane line between the horizon position and the hood position to obtain a first primary point crossing the horizon position and a second and a third primary point crossing the hood position; a step of accumulating and storing the first to the third primary point for a single photographed image; a step of calculating an average of a plurality of the first primary points accumulated and stored, an average of a plurality of the second primary points accumulated and stored, and an average of a plurality of the third primary points accumulated and stored if a number of times for accumulation and storage equals a prescribed number; a step of calculating a first distance between the average of the first primary points and the average of the second primary points and a second distance between the average of the first primary points and the average of the third primary points; and a step of determining a line between the average of the first primary points and the average of the second primary points and a line between the average of the first primary points and the average of the third primary points as lanes on both sides if the first and the second distance are shorter than or equal to a prescribed threshold.