Camera System for Vehicle Parking Using Homography
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle parking systems lack an effective method to provide drivers with an overhead view for parallel parking assistance using forward- and backward-facing cameras, relying on noisy sonar sensors or additional sideward-facing cameras.
Innovation Solution
A system utilizing a forward- and/or backward-facing camera to capture images of a parking area, creating a homography to estimate an overhead view, and displaying this view on a dashboard screen, with real-time updates of the vehicle's position relative to unoccupied parking spaces, leveraging existing camera technology for night vision and backover avoidance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If sonar sensors are used for parking assistance, then the system is inexpensive, but the measurements are noisy and inaccurate leading to false system behavior
Solution Approach 1:
The patent replaces sonar (acoustic) sensors with a camera-based vision system. The camera captures images that are processed to generate overhead views and detect parking spaces, eliminating the noisy and inaccurate measurements of sonar while maintaining cost-effectiveness through the use of standard camera hardware already present in vehicles.
Solution Approach 2:
The system creates a virtual copy or representation of the real-world parking environment by capturing images with the camera and generating a synthesized overhead view. This digital model allows for accurate detection and measurement without the physical limitations of sonar sensors.
2Measurement precision
If additional sideward-facing cameras are added for motion-stereo parking assistance, then parking area detection is improved, but device complexity increases
Solution Approach 1:
The patent makes the existing forward- and backward-facing cameras perform multiple functions: they not only provide their original functions (night vision, lane detection, backover avoidance) but also enable overhead view generation and parking space detection through image processing and homography transformations. This eliminates the need for additional sideward-facing cameras.
Solution Approach 2:
The system transforms 2D images captured by standard cameras into a synthesized 3D overhead view perspective through homography calculations. This dimensional transformation allows the existing cameras to provide information equivalent to what would require additional cameras positioned at different locations.
3Measurement precision
If feature-based maps are generated using vSLAM, then localization is achieved, but the maps are unusable by humans for localization
Solution Approach 1:
The system transforms the abstract feature-based map into a visually intuitive overhead view that resembles a real-world aerial perspective. By rendering parking spaces, vehicles, and surroundings in a familiar visual format with appropriate colors and perspectives, the map becomes immediately recognizable and usable by humans for localization and navigation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for assisting in parallel parking includes providing a vehicle with a forward-looking camera and/or a backward-looking camera. The camera is used to capture an image of a parking area including at least one unoccupied parking space and a plurality of parking spaces occupied by other vehicles. A homography of the captured image is created. The homography is used to estimate an image of the parking area from an overhead viewpoint. A portion of the overhead image including the unoccupied parking space is displayed on a display screen within a passenger compartment of the vehicle.