An Image-Based Method to Detect Steering Angle of Moving Vehicle

Junnan YANG,Zhongke SHI,Tian SHU
DOI: https://doi.org/10.7652/xjtuxb201306013
2013-01-01
Abstract:A novel image-based method is proposed to automatically detect steering angle for vehicle modeling and safety analysis. A CCD sensor is installed on vehicles to construct a visual detection system, and edge extraction is implemented on the image captured by the CCD sensor. The extracted edge points of the steering wheel are then used in an ellipse fitting, and a preliminary steering angle is calculated using the algorithm for detecting the rotation angle of circle. The continuity of the vehicle movement is used to analyze the effectiveness of the resulting preliminary steering angle. The statistical values of steering angle increments are used to modify the steering angle and to obtain the final steering angle of the vehicle. Experimental results show that the average error between values of angles measured by the proposed method and those measured by the steering wheel-based method is 0.25624°, while the average error between both of the angles is approximately 0.1° when the steering angles are in the concentrated interval [-4.15°, 4.15°]. The results show that the proposed method is able to effectively detect steering angles of moving vehicle during daytime.
What problem does this paper attempt to address?