A Vision-Based Unstructured Road Detection Algorithm for Self-driving Cars
01 Jan 2021-pp 369-379
TL;DR: In this paper, a novel vision-based road detection technique is proposed, which uses noise to enhance the road edges in the image and unstructured straight road is detected using Hough Transform.
Abstract: Unstructured road detection is one of the difficult tasks for self-driving cars than the detection of road with proper lane markings. Also, it is an extremely difficult task to detect the highly deteriorated district and taluk roads using currently available vision-based algorithm; as the exposed gravels and grass covering on both sides (edges) of road adds more noise in the input image. To address this issue, a novel vision-based road detection technique is proposed in this research work. This new method uses noise to enhance the road edges in the image and unstructured straight road is detected using Hough Transform. This paper is divided into three parts: bird’s eye view transformation of 2D road image received from the vehicle camera to correct the perspective distortion and easier identification of Region of Interest (ROI), addition of noise in the ROI of image to differentiate the valid road from the background and use of Hough Transform to identify the edges of unstructured road having no road markings. Finally, we present a simple way to find the centerline on the detected road for departure warning to reduce the additional computation. The simulation results corroborate that the proposed method detects the road successfully and can be used in real-time detection system.
References
More filters
••
TL;DR: This paper decomposes the road detection process into two steps: the estimation of the vanishing point associated with the main (straight) part of the road, followed by the segmentation of the corresponding road area based upon the detected vanishing point.
Abstract: Given a single image of an arbitrary road, that may not be well-paved, or have clearly delineated edges, or some a priori known color or texture distribution, is it possible for a computer to find this road? This paper addresses this question by decomposing the road detection process into two steps: the estimation of the vanishing point associated with the main (straight) part of the road, followed by the segmentation of the corresponding road area based upon the detected vanishing point. The main technical contributions of the proposed approach are a novel adaptive soft voting scheme based upon a local voting region using high-confidence voters, whose texture orientations are computed using Gabor filters, and a new vanishing-point-constrained edge detection technique for detecting road boundaries. The proposed method has been implemented, and experiments with 1003 general road images demonstrate that it is effective at detecting road regions in challenging conditions.
401 citations
••
06 Jun 2005TL;DR: A new method based on color information is proposed that is robust in the presence of noise, shadows, pavement, and obstacles such like cars, motorcycles and pedestrians conditions and this method is applicable in complex environment.
Abstract: Lane boundary detection is the problem of estimating the geometric structure of the lane boundaries of a road on the images captured by a camera. To be an intelligent vehicle, lane boundary is necessary information, so the system and the algorithm should be as simple and fast as possible. In this paper, we propose a new method based on color information and this method is applicable in complex environment. In this system, we first choose a region of interest to find out a threshold using statistical method in a color image. The threshold then will be used to distinguish possible lane boundary from the road. We use color-based segmentation to find out the lane boundary and use a quadratic function to approach it. This system demands low computational power and memory requirements, and is robust in the presence of noise, shadows, pavement, and obstacles such like cars, motorcycles and pedestrians conditions. The result images can be used as pre-processed images for lane tracking, road following or obstacle detection.
233 citations
••
05 Aug 1995TL;DR: AURORA is a vision-based system designed to warn a vehicle driver of possible impending roadway departure accidents, using a downward looking color video camera with a wide angle lens, a digitizer, and a portable Sun Sparc workstation.
Abstract: AURORA is a vision-based system designed to warn a vehicle driver of possible impending roadway departure accidents. It employs a downward looking color video camera with a wide angle lens, a digitizer, and a portable Sun Sparc workstation. Using a binormalized adjustable template correlation algorithm, it reliably detects lane markers on structured roads at 60 Hz. A time-to-lane-crossing (TLC) measurement is calculated for each image based on the estimation of vehicle's lateral position and velocity. This measurement is used to trigger an alarm when the TLC falls below a preset threshold. Promising results have been achieved under a variety of weather and lighting conditions, on many road types.
129 citations
••
TL;DR: A modified approach is proposed to accelerate the HT process in a computationally efficient manner, thereby making it suitable for real-time lane detection and lane departure prediction.
58 citations
••
TL;DR: This work introduces an approach to detect information about lane and vehicle for the driver assistance system, or the lane change assistant system, and combines the horizontal edge filter, the Otsu’s thresholding, and the vertical edge to detect vehicles.
43 citations