A robust lane detection and tracking method based on computer vision

Abstract

This paper presents a robust method designed to detect and track a road lane from images provided by an on-board monocular monochromatic camera. The proposed lane detection approach makes use of a deformable template model to the expected lane boundaries in the image, a maximum a posteriori formulation of the lane detection problem, and a Tabu search algorithm to maximize the posterior density. The model parameters completely determine the position of the host vehicle within the lane, its heading direction and the local structure of the lane ahead. Based on the lane detection result in the first frame of the image sequence, a particle filter, having multiple hypotheses capability and performing nonlinear filtering, is used to recursively estimate the lane shape and the vehicle position in the sequence of consecutive images. Experimental results reveal that the proposed lane detection and tracking method is robust against broken lane markings, curved lanes, shadows, strong distracting edges, and occlusions in the captured road images.

Computational Intelligence in Automotive Applications

“…Many feature-based algorithms are able to detect unmarked or unstructured roads based on the segmentationbased approaches [34]. Unlike the feature-based methods, model-based methods normally incorporate various constraints (such as a flat road or parallel road boundaries) during the parameter estimation stage to minimize the error and provide a simple description of the lane with mathematical models such as parabolas [14,35,18] or splines [30,32]. Existing parameter estimation algorithms include, among others, the Hough transform [18], active contour [32,12] and maximum a posterior estimation [14,35].…”

“…This step is often omitted in many existing lane analysis systems and only few tracking algorithms have been tested for this application [28,35,19]. However, the tracking step is very important since it gathers valuable information from previous results and utilizes this information to find the lane parameters quickly and accurately.…”

“…Some of the most important lane features are the edges [24,14,35,18,4,28,30,32,21] since the lane boundaries are normally assumed to be clearly marked with white or yellow paint in high contrast. It requires small computational power and defines sharp transitions in the image intensity.…”

IVVI 2.0: An intelligent vehicle based on computational perception

“…Thus, there are examples of road lane detection (Collado, Hilario, de la Escalera, & Armingol, 2008;Zhou, Xu, Hu, & Ye, 2006), and obstacles recog nition and avoidance in the vehicle's path such as either vehicles (Musleh, de la Escalera, & Armingol, 2012b) or pedestrians (Musleh, de la Escalera, & Armingol, 2011;Soquet, Perrollaz, Labayrade, & Auber, 2007) or other elements, like traffic lights and marks on roads (Franke et al, 2001). ADAS are on board vehi cle systems which focuses on the driving process.…”

On feature templates for Particle Filter based lane detection

“…Additionally, incorporating new cues to enhance the robustness of the estimation is straightforward. Recently, Zhou et al [17] proposed a maximum a posteriori formulation for the lane detection, applying a Particle Filter framework to simultaneously track lane shape and vehicle position. Franke et al [7] applied such a framework to the recognition of country roads, and Loose et al [13] for rural roads.…”