Research on image segmentation and feature extraction of freeway roadside landscape | IEEE Conference Publication | IEEE Xplore

Research on image segmentation and feature extraction of freeway roadside landscape


Abstract:

Landscape feature extraction and quantification have been a bottleneck problem when studying the relationship between roadside landscape and driving psychology. According...Show More

Abstract:

Landscape feature extraction and quantification have been a bottleneck problem when studying the relationship between roadside landscape and driving psychology. According to the diversification and detailed texture of roadside landscape unit, three kind methods such as Marker-based watershed algorithm, Canny edge detection algorithm and Texture analysis algorithm were explored for the research of freeway roadside landscape segmentation. The above approaches were tested by using experimental data obtained in Kunyuan freeway in Yunnan province. A contrastive analysis was conducted to identify the applicability of each algorithm and finally the texture algorithm was selected. The RGB color histograms and HSV average values were extracted from the segmented image, correlation between the two above features and the composition and continuity of roadside landscape were analyzed respectively, which can provide a new method to quantify freeway roadside landscape features.
Date of Conference: 29-31 July 2010
Date Added to IEEE Xplore: 20 September 2010
ISBN Information:

ISSN Information:

Conference Location: Beijing, China

1 INTRODUCTION

The extraction and quantification of freeway roadside landscape information are the bottlenecks in the research on its influence on driving psychology and behavior. Compared to the general image, the image of freeway roadside landscape is characteristic for the complexity of composing elements and variability with time and landform, so that the high-precision and high-robustness algorithm for image segmentation is needed. The freeway roadside landscape mainly consists of natural, cultural and ecological roadside landscape. The diversity of the composing elements lead to the following two marked characters: (a) The roadside landscape changes with regions, roadside slope and plant cover are large difference in different regions of China; (b) The roadside landscape changes with time, the color and plant cover are different in the same region in different times. In order to quantitatively research the influence of different roadside landscape on driving behavior, image processing technique is needed to automatically extract the characters of roadside landscape, and the elements of roadside landscape need to be expressed quantitatively, and then the relationship between driving behavior and characters of the roadside landscape can be established. The relationship among the above elements is shown as Fig. 1. Relationship among roadside landscape, image processing and driving behavior

References

References is not available for this document.