A Simple and Efficient Lane Detection using Clustering and Weighted Regression

Developing a vision based, efficient and automatic lane detection system from a moving vehicle is a challenging task mainly due to poor quality of lane markings, occlusion created by traffic, complex road geometry and nevertheless, the non-existence of unique models. The focus of this paper is to detect line or curve like segments from a video image taken from moving vehicle and merge them to detect road lane marks. In this work, we propose a method of applying weighted regression to fit a curve in place of traditional line grouping and fitting curve separately. The proposed vision based solution is simple but computationally efficient and hence works in real time on the image sequence captured by camera mounted in the vehicle.

• Type: Conference
• Year: 2009
• Where: COMAD
• Authors: Rudra N. Hota, Shahanaz Syed, Subhadip Bandyopadhyay, P. Radha Krishna
• Keywords: Automatic Lane Detection, COMAD 2008, COMAD 2009, Complex Road Geometry, Road Lane Marks