Rear Warning Apparatus Detecting Low-Height Curbs via Image Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle warning systems struggle to effectively alert drivers to hazards such as depressions or curbs, which are lower than the ground or have a relatively short height, as these objects are not adequately detected by traditional sensors and alarm devices.
Innovation Solution
A back warning apparatus and method that uses an image sensor and processor to capture and analyze rearview images, recognizing free space boundaries and surface profiles to determine the attributes of rear objects, and outputs warnings based on the vehicle's information and detected hazards, differentiating warnings based on the height of the vehicle and rear objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional sensors (ultrasonic, radar) are used for rear collision warning, then the system can detect objects at a distance, but it fails to detect objects with short height such as curbs and depressions
Solution Approach 1:
The patent introduces an image processing system as an intermediary between the rear camera and the warning output. This intermediary analyzes the captured image to identify free space boundaries and calculate depth information, enabling the system to detect objects that traditional sensors miss, such as curbs and depressions, by translating visual information into actionable depth data
Solution Approach 2:
The patent replaces the mechanical sensing approach (ultrasonic waves, radar) with an optical imaging approach. By using a rear camera to capture images and processing these images to extract depth information through free space boundary analysis, the system substitutes physical sensing mechanisms with optical-electronic processing, enabling detection of low-height objects
2Reliability
If the system provides comprehensive warnings for all rear objects, then safety is improved, but false alarms increase for non-hazardous objects
Solution Approach 1:
The patent applies local quality by providing differentiated warning outputs based on the specific characteristics of detected objects. Instead of a uniform warning for all objects, the system analyzes object attributes (height, depth, position) and generates appropriate warning levels, reducing false alarms for non-hazardous objects while maintaining high alerts for genuine threats
Solution Approach 2:
The patent changes the parameter of warning output based on the analyzed attributes of rear objects. By calculating depth information from free space boundaries and comparing object characteristics against hazard criteria, the system dynamically adjusts warning generation, ensuring reliable safety alerts while minimizing false alarms through parameter-based differentiation
3Measurement precision
If the system uses image processing to detect rear objects, then detection of low-height objects is improved, but processing time and computational complexity increase
Solution Approach 1:
The patent extracts only the essential information needed for hazard detection from the captured image - specifically the free space boundaries and depth calculations. By focusing on extracting these critical geometric features rather than processing the entire image comprehensively, the system achieves accurate detection of low-height objects while minimizing processing time
Solution Approach 2:
The patent applies partial action by performing image processing only on relevant regions - specifically analyzing free space boundaries where depth information is most critical for detecting rear hazards. This selective processing approach maintains high detection accuracy for low-height objects while reducing overall computational burden and processing time
Data Source
AI summary
Provided is a back warning apparatus and a method and control system therefor. The back warning apparatus includes an image sensor installed at a rear of a vehicle, and a controller that obtains a rearview image via the image sensor when the vehicle moves rearward, recognizes the boundary of a free space in the rearview image and generates a surface profile associated with the height/depth of a surface behind the vehicle, determines the attribute of a rear object based on the boundary of the free space and the surface profile, and outputs a warning by determining whether a hazard exists in association with the attribute of the rear object based on the information associated with the vehicle. According to the present disclosure, information associated with a rear object having a short height is provided, thereby making effective use of a parking space and preventing damage to the bodywork of the vehicle.