Deep learning based lane line detecting and early-warning device and method

Abstract

The invention relates to a deep learning based lane line detecting and early-warning device and method. The method comprises the following steps: acquiring an original lane image in the vehicle running direction; splicing the images and changing vision angles to obtain a complete front lane image; extracting the classifying property information of each pixel in the front lane image based on a pre-trained deep learning model; determining that a pixel point belongs to the lane line or a non-lane line based on the pixel classifying property information; analyzing the obtained lane line information through the lane deviation analyzing method; and determining the intensity of the sent early-warning information. With the adoption of the method, a current lane line detection method can be optimized; the problem that the vision angles in image acquiring influence the detection of the actual lane line and the lane line curvature to be computed in the current lane line detection method can be solved; the recognizing accuracy and robustness are improved; and the driving safety is improved.

Description

technical field [0001] The invention belongs to the technical field of automobile traffic safety, and in particular relates to a lane line detection and early warning device and early warning method based on deep learning, which is suitable for lane line detection and early warning under complex road conditions. Background technique [0002] Lane line detection and early warning refers to sensing the lane through on-board sensors, judging the position of the vehicle within the lane line, analyzing whether the vehicle is in danger of deviating from the lane line, and issuing a departure warning. In the prior art, the lane line detection and departure warning system generally processes the image information collected by the camera in real time, extracts the features of the lane line through image filtering technology, obtains the parameter value of the straight line where the lane line is located by using Hough transform, and then estimates the parameters. Estimate the paramet...

AI Technical Summary

Problems solved by technology

The effect of lane line detection under complex road conditions is poor, and the angle of view of the image collected by the general vehicle camera is basically the same as that of the driver. It does not consider the impact of the angle of view of the collected image on the actual lane line detection and the lane that may need to be calculated. The problem of line curvature, when analyzing and judging the lane departure, did not consider adjusting the strength of the generated warning signal according to the actual degree of deviation

Images