Autonomous vehicles and modern Advanced Driver Assistance Systems require a module for detecting parking lot markings, in order to locate free parking slots and to assist the driver with the parking maneuver or to entirely accomplish it. Equipping those systems with multiple fisheye cameras is a common design solution, since fisheye sensors provide a wide field of view at a cost and size comparable with rectilinear cameras. Unfortunately, a straight line in the world corresponds to a segment of conic section in fisheye images, hence the detection of parking lot markings have to deal with non trivial problems. This paper proposes a method that implements the Progressive Probabilistic Hough Transform in the space obtained from the reprojection of the fisheye image points in the equivalence sphere. The presented solution allows to identify road marking segments directly on the fisheye image, providing long range accuracy un-like typical bird's-eye view based approaches. A simple parking slots detector based on the detected road markings has been developed in order to verify the feasibility of such a system. Finally, the intrinsic iterative nature of the proposed algorithm makes it suitable for real-time computation, as required by automotive scenarios.
Parking slots detection on the equivalence sphere with a progressive probabilistic hough transform
2017-10-01
966670 byte
Conference paper
Electronic Resource
English
Robust Detection of Lines Using the Progressive Probabilistic Hough Transform
British Library Online Contents | 2000
|3D Hough transform for sphere recognition on point clouds
British Library Online Contents | 2014
|Hough Transform for Parking Space Estimation Using Long Range Ultrasonic Sensors
SAE Technical Papers | 2006
|Hough transform for parking space estimation using long range ultrasonic sensors
Automotive engineering | 2006
|A LiDAR-Based Parking Slots Detection System
Springer Verlag | 2024
|