Stereo Vision-Based Detection Method of Drivable Area of ​​Intelligent Vehicles

Abstract

The invention discloses a method for detecting a drivable region of an intelligent vehicle based on stereo vision. Based on stereo vision technology, the parallax value of an input image is calculated, and at the same time, parallax confidence is introduced to eliminate wrong parallax values, and then the original image is horizon-lined. Detect to narrow the target search space, then calculate the position probability of obstacles on the road surface, and finally use the dynamic programming optimization solution method to obtain the driving area of ​​the intelligent vehicle. The present invention introduces disparity confidence to eliminate false disparity, and designs a pixel-level obstacle position probability model. This method can meet the real-time requirements, and can effectively detect small and distant obstacles, greatly improving the detection accuracy. It has good environmental robustness, is suitable for the detection of drivable areas in complex traffic environments, and has very important practical value and engineering application prospects.

Description

technical field [0001] The invention belongs to the field of intelligent transportation, and relates to a method for detecting a drivable area of ​​an intelligent vehicle based on stereo vision. Background technique [0002] With the continuous progress and development of science and technology, there are more and more vehicles and the traffic environment is becoming more and more complex. Many auxiliary driving systems not only make people's driving experience simple and convenient, but also ensure people's traffic safety and reduce the incidence of car accidents. In recent years, the research on intelligent vehicles at home and abroad is very hot. [0003] Intelligent vehicle drivable area detection is an important part of the intelligent driving platform, which has important research significance and application value. [0004] Although a large number of scientific researchers have invested in the research of intelligent vehicle drivable area detection technology, there ...

AI Technical Summary

Problems solved by technology

The natural weather outside the vehicle is complex and changeable, and the extremely harsh natural weather with low visibility, such as rain, fog, snow, smog, etc., as well as the dark night environment with poor lighting conditions, all cause great damage to the data collection work of sensors, especially visual sensors. big problem

The third is the impact of overly complex traffic scenes

The biggest advantage of the former is that the calculation is simple and it is easy to ensure real-time performance. However, this type of method is more dependent on the quality of the disparity map, and when the scene is too complex and there are many obstacles and irregular shapes, it is difficult to extract the straight line of the V disparity map. As a result, the obtained obstacle position is wrong, and the detection accuracy of the drivable area is reduced.

The latter is more robust to the environment, and can have better detection results in harsh environments such as nights with low visibility, rain, fog, and snow. However, the biggest problem with this type of method is that the edge information of the left and right views is very complicated. Mis-match is prone to occur when matching left and right views, which leads to the problem of low detection accuracy of drivable areas in this method when there are too many obstacles in the scene

Images