Year: 2021
Type: article
Abstract: Vision-based road detection is a challenging topic in the field of autonomous vehicles. Road information is needed for a robot vehicle to control their movement while driving. The aim of this study is to make a simple and fast road detection system which can be used to control vehicle robots in real-time. This paper proposes real-time road detection, which begins from images captured on a camera, then pre-processing and road detection based on color and statistical analysis, for reducing the detection failure rate. The dataset used for the system was taken from urban roads. The color-matching procedure uses minimum/maximum matching between the dataset and sample pixels of captured images. At the same time, the captured images are clustered by a grid method, grouping pixels having the same size between grid cluster and dataset. PSNR and MSSIM are used for analyzing the similarity of the grid cluster and dataset. We implemented a threshold value to determine if the PSNR and MSSIM analysis result is a road. From our experiment, we obtain a threshold value of 85 db for PSNR and 0.90 for MSSIM. If the both analyses result in values greater than their threshold values, then the result is voted to be a road. Finally, the road is determined by analyzing the result of color and statistical analysis. The system will determine there is a road if the both analyses detect the road. The result shows the color based road detection resulting file detection caused by changing the lighting and the color of the road. We have solved that problem by implementing PSNR and MSSIM analysis in the grid clustering method. The result of the proposed method has been presented in this paper, and can potentially reduce the failure rate of detection using the color-based method. Keywords: road labeling, image measurement, mean structural similarity index measurement.
Language: English
Subfield: Automotive Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W3135358874
