01661283

Component

From the driver's visual perception, this paper studies the video image of the on-vehicle recorder. Combining global threshold with local threshold,a segmentation method of gray image is proposed.80 percent road environment images of are accurately divided into parts: sky, roadway and landscape.68 parameters are extracted for every image to represent the attributes of road environment, including area parameters, shape parameters, color parameters, and texture parameters of landscape. Statistical analysis methods are used to develop the relationship between these parameters and speed. The result of multiple linear regression shows that there is a significant linear correlation between speed and 14 parameters, in which 99 parameters belong to landscape and belong to roadway and 1 belongs to sky. The relationship proves that the parameters of landscape, roadway and sky have influence on speed, and the parameters of landscape have a greater effect than road and sky. Factor analysis and clustering are used to classify road environment into 5 categories, in which category 5 has the largest mean and the smallest standard deviation of speed and is considered the best type of 5 kinds of road environment.

This paper was sponsored by TRB committee AFB40 Standing Committee on Landscape and Environmental Design.

18-01691

English

21p

2018

Transportation Research Board 97th Annual Meeting

Digital/other

Figures; Photos; References; Tables

Cluster analysis; Factor analysis; Image processing; Linear regression analysis; Roads; Speed; Visual perception

Environment; Highways; Safety and Human Factors

Transportation Research Board Annual Meeting 2018 Paper #18-01691

TRIS, TRB, ATRI

Jan 8 2018 10:25AM