Embedded Files
Road (lane) detection for UGV by a stereo camera
Road (lane) detection for UGV by a stereo camera
In this project, we aim at finding traversable road region for UGV based on a stereo camera. First, we estimate a rough region by stereo disparity map. With the horizon and flatness constraints, we can robustly locate the vanishing point of road region. Second, a weighted graph is constructed using all pixels of the image, and the detected vanishing point is treated as the source node of the graph. By computing a vanishing-point constrained Dijkstra minimum-cost map, where both disparity and gradient of gray image are used to calculate cost between two neighbor pixels, the problem of detecting road borders in image is transformed into that of finding two shortest paths that originate from the vanishing point to two pixels in the last row of image. The proposed approach has been implemented and tested over 6000 grayscale images of different road scenes in the KITTI dataset and off-road scenarios.
Referred paper:
Yigong Zhang, Yingna Su, Jian Yang, Jean Ponce, and Hui Kong: When Dijkstra meets vanishing point: a stereo vision approach for road detection, IEEE Transactions on Image Processing, 2018
Yingna Su, Yigong Zhang, Tao Lu, Jian Yang, and Hui Kong: Vanishing point constrained lane detection with a stereo camera, IEEE Transactions on Intelligent Transportation Systems, 2017
Sainan Guo, Yicheng Gao, Yingna Su, and Hui Kong, Illumination-invariant Feature by SVD and its Applications , ACPR 2017
Tao Lu, Yigong Zhang, and Hui Kong, Detection of Road Vanishing Point by Stereo Camera, ACPR 2017
2D Lidar + single camera: urban road detection for UGV
2D Lidar + single camera: urban road detection for UGV
In this project, we propose a sensor fusion strategy for urban road detection. The sensors include a monocular camera and a 2D single-line Lidar scanner. This sensor fusing method has been tested on over 4000 images (including the KITTI and Oxford robotcar datasets). It can achieve very promising performance comparable to the state-of-the-arts road detection methods (including deep learning methods). This method is training free, i.e., we do not need collect any road image sample to learn a classifier.
Referred paper:
Yingna Su, Yigong Zhang, Jose M. Alvarez, and Hui Kong, An Illumination-invariant Nonparametric Model for Urban Road Detection Using Monocular Camera and Single-line Lidar, IEEE ROBIO 2017, (oral presentation)
Vanishing point detection for road detection
Vanishing point detection for road detection
In this project, we propose a general single-camera basaed framework for road detection with the constraint of vanishing point of road region. This method can find vanishing point of road region under different conditions. Therefore, this method can be used to detect different types of road (including unstructured roads in rural areas or desert).
Referred papers:
Hui Kong, Sanjay E. Sarma, Feng Tang, Generalizing Laplacian of Gaussian Filters for Vanishing-Point Detection, IEEE Transactions on Intelligent Transportation Systems, 2013
Hui Kong, Jean-Yves Audibert, and Jean Ponce, General road detection from a single image, IEEE Transactions on Image Processing, 2010
Hui Kong, Jean-Yves Audibert, and Jean Ponce, Vanishing point detection for road detection, IEEE CVPR, 2009
Google Sites
Report abuse