Adaptive lane line detection and warning algorithm based on dynamic constraint
- PDF / 2,645,652 Bytes
- 17 Pages / 439.37 x 666.142 pts Page_size
- 22 Downloads / 188 Views
Adaptive lane line detection and warning algorithm based on dynamic constraint Jinliang Gong 1 & Yanfei Zhang 2 & Ke Sun 1 & Xiaofeng Sun 3 Received: 22 April 2020 / Revised: 14 July 2020 / Accepted: 28 July 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
The lane line detection algorithm is highly sensitive to the environment. And the selection of parameters is greatly affected by human subjective factors. Therefore, the concept of adaptive and dynamic constraint is introduced into the lane line detection algorithm. The evaluation function is set based on the detected information. A high precision lane line detection algorithm with supervision and warning mechanism is proposed. Firstly, the Canny operator gradient calculation method is improved by using eight neighborhood models. And by setting multilevel threshold instead of double threshold detection method, the algorithm’s adaptability is improved. Secondly, the concept of dynamic region of interest is proposed. The detection region of Canny operator is constrained, the environment interference to the algorithm is reduced. Finally, lane line information is included in the supervision and correction warning mechanism. And the early warning is given to drivers. The experimental results show that the algorithm is of high accuracy, real-time and adaptability. Keywords Lane line detection, adaptive threshold, canny operator, dynamic constraints, departure pre-warning
1 Introduction Self-driving cars are complicated hardware complex that combines a variety of cutting-edge science and technology [4, 12, 19, 24]. Computer Vision, Ssensor Fusion, Machine Learning and other Chinese Library Classification: TP391.4 Document code: A
* Yanfei Zhang [email protected]
1
School of Mechanical Engineering, Shandong University of Technology, Zhangdian District, Zibo 255049, China
2
School of Agricultural Engineering and Food Science, Shandong University of Technology, Zhangdian District, Zibo 255049, China
3
Goertek Technology Limited Company, Qingdao 266000 Shandong, China
Multimedia Tools and Applications
emerging technologies ensure the safety of self-driving cars [11, 20–22]. Among them, Computer Vision is the key technology for road recognition and detection in self-driving. The identification of structured roads is mostly based on the inherent properties of images such as color and edge currently [1, 10, 15]. Wenjie Song et al. [17] proposed a light driving lane detection and classification system based on stereo vision to realize the ego-car’s lateral positioning and forward collision warning to help advanced driver assistance systems (ADAS). Duan et al. [6] has proposed road and lane line extraction algorithm based on Image Segmentation and Voting Function, and the algorithm uses K-means clustering algorithm to segment images according to color under the premise of constant illumination. Cai Yingfeng et al. [3] has detected core segments in an image with a modified Canny operator, and extracted linear targets in segments. Sebdan
Data Loading...
