Road environment modeling using robust perspective analysis and recursive Bayesian segmentation

Abstract

Recently, vision-based advanced driver-assistance systems (ADAS) have received a new increased interest to enhance driving safety. In particular, due to its high performance-cost ratio, mono-camera systems are arising as the main focus of this field of work. In this paper we present a novel on-board road modeling and vehicle detection system, which is a part of the result of the European I-WAY project. The system relies on a robust estimation of the perspective of the scene, which adapts to the dynamics of the vehicle and generates a stabilized rectified image of the road plane. This rectified plane is used by a recursive Bayesian classifier, which classifies pixels as belonging to different classes corresponding to the elements of interest of the scenario. This stage works as an intermediate layer that isolates subsequent modules since it absorbs the inherent variability of the scene. The system has been tested on-road, in different scenarios, including varied illumination and adverse weather conditions, and the results have been proved to be remarkable even for such complex scenarios.

Gömülü bir platform üzerinde gerçek zamanlı şeritten ayrılma uyarı sistemi

“…Şerit tespit işlemlerinin bu gibi olumsuz durumlardan daha az etkilenmesini sağlamak amacıyla Nieto vd. 'nin [21] çalışmasında kullanılan ve Eş. 1'de verilen şerit işareti filtresi kullanılmaktadır.…”

Lane Detection and Tracking Using a Parallel-snake Approach

“…f y (x, y) = k bal · n(s) − ∂E ext ∂y (22) where k bal is the weighted factor of the balloon force to n(s).…”

“…Although perspective distortion was eliminated in [5,11,15,22,23], the parallel property of the two boundaries was considered only in [11,22]. The least squares method was used to fit lane models in [11,22], while the Bsnake method [27,28] was proposed to estimate lane boundaries through iteration. Compared with other models, the B-snake model is robust in the presence of shadows because the external force of the snake is based on the gradient of the image instead of individual features.…”

Monovision‐based vehicle detection, distance and relative speed measurement in urban traffic

“…This method could easily overestimate the measure due to the difficulty of getting the exact tire-asphalt contact point which in addition does not correspond to the real back of the vehicle. [17] uses known information about the vehicle appearance (the vehicle's underside) and the distance is determined using the inverse perspective transform assuming flat earth. Finally, [18] assumes an ideal vehicle width for all vehicles and uses the camera pinhole model to determine the distance of the vehicle.…”

“…Finally, as the shadow is on the road plane and assuming flat earth as in [17], a first rough estimation of the vehicle's distance is obtained based on the location of the lower edge of the vehicle's bounding box (the shadow's lower edge) in the image. This approximate distance is very useful because it in turn provides values of the vehicle number plate dimensions at this distance which are exploited in the number plate detection algorithm (section 3.5).…”