Citation

Martin, Aerun and Kamaruddin, Mohd Nazeri and Md Sani, Zamani (2022) Shadow detection and removal of road images with maximum extraction of road markers using multi-colour-space. In: 2nd FET PG Engineering Colloquium Proceedings 2022, 1-15 December 2022, Multimedia University, Malaysia. (Unpublished)

Text 11_Aerun_PPM Extended Abstract_Second Half 2022.pdf Restricted to Registered users only Download (69kB)

Abstract

Background - Advanced Driver Assistance System (ADAS) provides automated assistance for safe vehicle manoeuvring on the road to reduce road traffic crashes. The decision-making capability of the system relies on the information feed in or collected during vehicle movement, such as road markers. Road markers provide essential information to the system, such as the number of lanes, path trajectory, and possibility of overtaking a vehicle. However, illumination conditions, especially shadows, are a significant challenge for abstracting the correct information from road markers, which could lead to detection errors. Purpose – This paper proposed a methodology to eliminate shadows on road markers by channel averaging from three channels, each from different colour spaces, Saturation, S from HSV, Blue-difference, Cb from YCbCr and Blue/Yellow value, and B from LAB. Design/methodology/approach – The three-colour channel averaging provides high sensitivity to the shadowed region and increases the distinguishment of shadowed and non-shadowed pixels for shadow elimination. However, this process also classified road marker pixels with the shadow being classified as the shadowed region. Thus, adaptive thresholding is applied to the shadow-classified pixels to detect pixels with road markers and shadows. Then, all the road marker pixels were grouped and converted to binary pixels white for feature extraction. Findings – This method can retain more than 98% of the road marker information. The shadow detection and elimination of road markers while retaining road marker information was tested on Caltech Dataset, and the validation of the method shows 98% of accuracy. Research limitations– The experiment is only conducted on the Caltech dataset, which consists of dataset road images captured on a sunny day with shadows by the trees on the side of the roads. The occultations and the efficiency of other datasets are not identified. Originality/value – The shadow detection methodology using three colour spaces was loosely based on previous research on hybrid colour space performance.

Item Type:	Conference or Workshop Item (Other)
Uncontrolled Keywords:	Shadow detection, Road marker, Channel- Averaging, Multi-Colour Space
Subjects:	T Technology > TA Engineering (General). Civil engineering (General)
Divisions:	Faculty of Engineering and Technology (FET)
Depositing User:	Ms Nurul Iqtiani Ahmad
Date Deposited:	16 Feb 2023 06:06
Last Modified:	16 Feb 2023 06:06
URII:	http://shdl.mmu.edu.my/id/eprint/10822

Downloads

Edit (login required)