Cited time in webofscience Cited time in scopus

Improvement of stereo vision-based position and velocity estimation and tracking using a stripe-based disparity estimation and inverse perspective map-based extended Kalman filter

Title
Improvement of stereo vision-based position and velocity estimation and tracking using a stripe-based disparity estimation and inverse perspective map-based extended Kalman filter
Author(s)
Lim, Young-ChulLee, MinhoLee, Chung-HeeKwon, SoonLee, Jong-hun
Issued Date
2010-09
Citation
Optics and Lasers in Engineering, v.48, no.9, pp.859 - 868
Type
Article
Author Keywords
Disparity estimationSub-pixel interpolationPosition and velocity estimationInverse perspective mapExtended Kalman filter
Keywords
Calibration ProcessDETECTION SystemDisparity EstimationDisparity EstimationsError CovariancesEstimationExtended Kalman FilterExtended Kalman FiltersFuzzy ControlImage CodingInterpolationInverse Perspective MapKeypointsLong RangeMatching ErrorMeasurement ErrorsMeasurement ModelMeasurement SystemMoving ObstaclesMoving VehiclesNearest PointObstacle DetectionOptimal PerformanceOptimizationPixelsPosition and Velocity EstimationPosition ControlPROPAGATIONQuantization ErrorsReal Road EnvironmentsRelative VelocityStable ErrorStereo VisionStereo Vision SystemSub-Pixel DisparitySub-Pixel InterpolationVehiclesVelocityVertical EdgesZero-Mean Normalized Cross Correlations
ISSN
0143-8166
Abstract
This paper presents a method for estimating the position and velocity of a moving obstacle in a moving vehicle. In most stereo vision systems, an obstacle's position is calculated by triangulation or an inverse perspective map (IPM) approach. However, measurement errors increase at long range due to quantization errors and matching errors. The key point that reduces measurement errors is to estimate the disparity accurately and precisely. This article focuses on the improvement in precision because accuracy can be enhanced through a calibration process in most measurement systems. The proposed method has two steps. One is to estimate sub-pixel disparities using a stripe-based accurate disparity (S-BAD) method. The other is to estimate and track the position and velocity of the obstacle with an IPM-based extended Kalman filter (EKF). The S-BAD method estimates accurate sub-pixel disparities with stripe-based zero-mean normalized cross correlation (ZNCC) using the vertical edge features within the dominant maximum disparity region that correspond to the nearest points from the host vehicles. The S-BAD method has the advantage of minimizing the quantization error and matching ambiguity and enhancing the precision of the disparity for the obstacle. The IPM-based EKF minimizes the error covariance and estimates the relative velocity while predicting and updating the state of the obstacle recursively. The method also gives optimal performance due to the measurement model with the stable error covariance. The experimental results show that the S-BAD method improves the precision of estimating the distance, and the IPM-based EKF minimizes the error covariance of the velocity in real road environments. © 2010 Elsevier Ltd. All rights reserved.
URI
http://hdl.handle.net/20.500.11750/3690
DOI
10.1016/j.optlaseng.2010.04.001
Publisher
Elsevier Ltd
Related Researcher
Files in This Item:

There are no files associated with this item.

Appears in Collections:
Division of Automotive Technology Advanced Radar Tech. Lab 1. Journal Articles
Division of AI, Big data and Block chain 1. Journal Articles
Division of Automotive Technology 1. Journal Articles

qrcode

  • twitter
  • facebook
  • mendeley

Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.

BROWSE