Light-stripe-projection-based target position designation for intelligent parking-assist system

This paper proposes a novel light-stripe-projection-based target position-designation method for an intelligent parking assist system, providing an economical free-space-based target position-designation method for poorly lit indoor parking spaces without degradation of the external appearance when integrated into a vehicle's body. The proposed method can be applied by adding a low-cost light plane projector to existing backward-parking cameras. Light stripe features are detected by the difference between an image with the light plane projector on and off. Three-dimensional information of the parking area is reconstructed using the light-stripe-projection theory. By orientation normalization and depth-map construction, free spaces are detected, and the nearest is selected as the target position. Experimental results show that the proposed method can successfully designate the target position, in spite of poor lighting conditions and the black reflective surface of adjacently parked vehicles.