T. Lin, S. Patnaik

Pages: 23-32

Abstract
Aiming at the problems of low warning accuracy and long warning time in traditional underground parking lot parking safety early warning methods, the application of multiple vision sensors in underground parking safety warnings is studied. The parking space is measured by multi-vision sensors, the width of the parking space is obtained, the parking space of the underground parking lot is detected based on the registration algorithm, the attitude angle of the obstacle vehicle is obtained, and the parking width and the attitude angle of the obstacle vehicle are denoised and corrected. The corrected multi-vision sensor detects the data, constructs two types of parking space models, estimates the vehicle pose, constructs the underground parking lot parking risk assessment model, sets a fixed threshold, and makes early warning judgments on the parking status to complete the underground parking safety warning in the parking lot. The results of simulation experiments show that the proposed method has higher accuracy and shorter warning time for parking safety in underground parking lots.
Keywords: multi-vision sensor; underground parking lot; parking safety risk warning; registration algorithm; vehicle attitude