Generative AI AVM Calibration for Marker-Based Distortion Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AVM systems face challenges in maintaining high-quality video quality due to changes in vehicle conditions and improper calibration by unskilled personnel, especially in non-dedicated facilities and varying lighting environments.
Innovation Solution
A method using generative artificial intelligence (AI) with a policy network, value network, and control network for marker-based data processing to calibrate camera settings, involving full-automatic and semi-automatic modes for accurate AVM video distortion removal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If factory calibration with dedicated facilities and skilled personnel is used, then AVM measurement precision is improved, but device complexity and calibration cost increase
Solution Approach 1:
The system performs self-calibration using the vehicle's own camera and processing unit. The calibration device captures images of markers, extracts feature points, and calculates camera parameters automatically without requiring external skilled personnel or complex dedicated facilities, thus resolving the contradiction between measurement precision and device complexity
Solution Approach 2:
The patent replaces the mechanical/physical calibration system (dedicated facilities, positioning machinery) with a computational system using image processing and algorithms. The camera parameter calculation is performed through software-based feature point extraction and coordinate transformation rather than physical measurement tools, reducing device complexity while maintaining precision
2Manufacturing precision
If factory calibration is performed, then initial AVM video quality is improved, but calibration reliability deteriorates when vehicle conditions change
Solution Approach 1:
The calibration system is designed to be dynamic and adaptable to changing vehicle conditions. It can perform recalibration at different stages (factory, repair shop, or by end users) when vehicle conditions change due to cargo weight, tire pressure, or part replacement, thereby maintaining calibration reliability and AVM video quality throughout the vehicle's operational life
Solution Approach 2:
The system allows recalibration of camera parameters (installation position, orientation, angle of view, distortion coefficients) when vehicle conditions change. By detecting feature points of markers under different conditions and recalculating parameters, the system adapts to parameter changes caused by vehicle modifications, maintaining reliable AVM performance
3Ease of operation
If repair shop calibration is attempted, then accessibility is improved, but measurement precision deteriorates due to lighting and marker distortion
Solution Approach 1:
The patent introduces a calibration device as an intermediary that captures images of the markers and performs automated feature point extraction. This intermediary system bridges the gap between simple accessibility and precise measurement by using image processing algorithms to accurately detect marker features even in non-ideal lighting conditions, enabling repair shops to perform reliable calibration without requiring controlled environments
4Ease of operation
If unskilled personnel perform calibration, then operational simplicity is improved, but calibration precision deteriorates
Solution Approach 1:
The calibration system is designed for self-service operation where unskilled personnel can perform calibration by simply placing markers and capturing images. The automated feature point extraction and parameter calculation algorithms handle the complex processing, enabling accurate calibration without requiring skilled personnel while maintaining high precision through computational methods
Data Source
AI summary
The present invention generally relates to a technology that calibrates a camera setting for a vehicle AVM system. In particular, the present invention relates to an AVM calibration technology by use of generative artificial intelligence (AI) that searches for a camera parameter for removing AVM video distortion by performing marker-based data processing on a video imaged by a camera of a vehicle AVM system by use of a generative AI model including a policy network, a value network, and a control network. The present invention is advantageous in that an AVM system can be stably calibrated even without a dedicated calibration facility and skilled personnel. In addition, the present invention is advantageous in that, even in a case where full-automatic-mode calibration fails, AVM calibration can be performed with reduced user intervention compared to that in the related art, by performing semi-automatic-mode AVM calibration using outline information of a marker.