Road and vehicle detection in highway scene for automotive FMCW antenna array radar

This paper deals with the road and vehicle detection problem in highway scene for automotive radar which uses frequency-modulated continuous-wave (FMCW) signals and antenna array. Due to the automotive radar´s low angle resolution, targets such as guardrails on road edge interfere with the vehicle detection and bring difficulties to the distance-velocity decoupling. We proposed an approach with Hough transformation and image matching for the road and vehicle detection. The interference of outroad targets is suppressed and the decoupling is achieved with low complexity. Specially, based on the road boundary model, the road boundary can be achieved by a curve Hough transform. Then, the on-road vehicles are detected and the distance-velocity decoupling can be solved by target region image matching with a total variation minimization method. Our new algorithm has a low computational complexity, and is validated by field experimental results.