Line-structured light pavement detection has been developed for about two decades, but it is not yet mature because of unstable measurement results. An image-processing procedure based on ROI (region of interest) images is proposed in this study, which includes slight center extraction, connection, smoothing, and the rutting parameter extraction. The light center extraction is the foundation of the entire process. Two extraction methods are discussed in this study, Gaussian method and the gravity method. The Gaussian method can find the light center curve precisely, but the calculation process is time consuming. The gravity method is based on the assumption that the light intensity of the laser line distributes uniformly along the cross-sections, which is a simplification of the Gaussian method and has a high processing efficiency. The light center curve extracted by Gaussian method is considered as the true value to be used to evaluate the accuracy of the gravity method. The result shows that the error introduced by simplification is acceptable, and the gravity method can be used in practice instead of the Gaussian method.
[1]
Peter M. Will,et al.
Grid Coding: A Preprocessing Technique for Robot and Machine Vision
,
1971,
IJCAI.
[2]
K H McGhee,et al.
AUTOMATED PAVEMENT DISTRESS COLLECTION TECHNIQUES
,
2004
.
[3]
Zhang Guangjun,et al.
Study on method for processing image of strip in structured-light 3D vision measuring technique
,
2003
.
[4]
Lou Xiaoping.
Extraction algorithm of light stripes center in the measurement system of structured light
,
2009
.
[5]
N. Otsu.
A threshold selection method from gray level histograms
,
1979
.
[6]
Makoto Nagao,et al.
Automatic Pavement-Distress-Survey System
,
1990
.