Multi-feature fused lane line detecting system and method and advanced driving assistance system

Abstract

The invention discloses a multi-feature fused lane line detecting system and method and an advanced driving assistance system. The method comprises the following steps that a top view of a lane area is captured; a color feature graph is obtained by extraction according to colors of lane lines and road interference items in the driving process of a vehicle; a contrast feature graph is obtained according to pixels in the color feature graph; a linear line is detected in the contrast feature graph according to projection, and an interference item in the linear line is rejected via geometric characteristic of the lane line; and the correct lane line is obtained by detection. According to the invention, color information is introduced into lane line detection for the first time, and anti-interference capability for light at night can be enhanced. The contrast feature graph which is obtained on the basis of a lane line enhancing operator of the pavement contrast is redefined. A geometric change rule of the lane line serves as a significant lane line rejection feature, and the interference items are removed from the linear line via geometric feature rejection of the lane line.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a multi-feature fusion lane line detection system, method and advanced driving assistance system. Background technique [0002] For advanced driver assistance systems (ADAS) based on vision algorithms, lane line detection is one of the key technologies. Lane line detection methods can generally be divided into methods based on line detection and methods based on key point detection. [0003] 1) The method based on line detection generally preprocesses the grayscale image through edge detection to obtain the edge enhancement map. On this basis, the Hough transform is used to detect straight lines, and finally the interfering straight lines are rejected through certain rules. 2) The method based on key point detection generally also filters the grayscale image through edge detection, and then extracts several edge points as key points. Finally, a number of lines are fitted as lan...

AI Technical Summary

Problems solved by technology

But in fact, different colors of light will have an impact on lane line detection

Especially at night, the light from taillights and traffic lights will affect the accuracy of lane line detection

Pure grayscale images cannot filter out these disturbances

[0005] In addition, whether it is based on the line detection method or the key point detection method, it mainly uses the strong edge characteristics of the lane line for detection, and the features used are relatively single, and it lacks the utilization of the change law of the lane line in the time domain, and lacks Combine multiple features organically

At the same time, it will cause more false detection or missed detection.

Images