Ternary tree traversal-based lane line detection method

Abstract

The invention discloses a ternary tree traversal-based lane line detection method and relates to the field of image processing. According to the method, candidate feature points which represent the center line of a lane line are extracted, and the number of pixel points required to be processed is reduced. The calculation efficiency of the algorithm is improved. The feature points of lane lines and noises are analyzed, and a method for filtering pseudo feature points irrelevant to the parameter space is provided. In this way, the parameter space conversion and the complex mathematical operation are avoided. The accuracy of lane line detection is improved. Meanwhile, a lane line prediction model is provided for carrying out lane line prediction in special road conditions. The internal memory is stored in the model. According to the change tendency of an image sequence from the past to the current, a lane line in a special road condition can be predicted. The lane line is predicted properly and reasonably. The accuracy of the detection result is improved.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a lane line detection method based on ternary tree traversal. Background technique [0002] The introduction of applications such as road sign detection, road lane line detection, and lane departure warning has made the development of vehicle intelligent systems more rapid. Road lane detection is a key technology in the vehicle intelligent system architecture. Because of the variability and uncertainty of road scenes, lane line detection has become a challenging topic. In the lane detection technology, real-time performance and accuracy are the main research goals, and also the main indicators reflecting the overall performance of a system. [0003] Among many road lane detection methods, the method based on machine vision has become the mainstream development trend. According to its detection principle, it can be roughly divided into four categories: methods based on Hough tra...

AI Technical Summary

Problems solved by technology

[0004] The improved method based on the Hough transform has a high misjudgment rate and computational complexity. When the road condition information is complex, the classic Hough transform is not feasible, and the detection effect of this type of method for curved roads is poor.

The detection range of the stereo vision-based method is very limited and the robustness is poor

Linear regression-based methods have low computational cost, but are very sensitive to noise, severely reducing the accuracy of detection algorithms

The method based on machine learning has been applied in some practical scenarios as a relatively good data classification tool. However, when the constructed data set is insufficient, its detection effect will be greatly reduced.

Images