Scene-Adaptive Camera Calibration for Image Distortion Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Distortion occurs in captured images due to impacts or deterioration of camera performance, causing subjects to be stored in shapes different from their actual appearance.
Innovation Solution
An electronic apparatus that extracts feature points from captured images based on scene types and corrects camera parameters using a processor to obtain and integrate calibration parameter values, updating the memory with these values to correct image distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If camera parameters are corrected using a single fixed method, then the correction process is simple, but the accuracy of distortion correction deteriorates under different scene conditions
Solution Approach 1:
The patent segments the correction process by dividing images into different scene types (outdoor, indoor, regular pattern) and applying different feature point extraction methods for each type. This segmentation allows the system to achieve high correction accuracy for each specific scene condition while maintaining a manageable overall process structure.
Solution Approach 2:
The patent changes the parameters of the feature point extraction method based on the identified scene type. By adjusting extraction parameters according to scene characteristics, the system optimizes correction accuracy for each scene type without requiring a completely different correction approach for each case.
2Ease of operation
If feature points are extracted using a generic method for all images, then the extraction process is straightforward, but the calibration accuracy deteriorates for specific scene types
Solution Approach 1:
The patent implements a dynamic feature extraction approach where the extraction method automatically adapts based on the detected scene type. The system transitions from a static generic extraction method to a dynamic scene-adaptive method, improving calibration accuracy while maintaining ease of operation through automated scene detection.
Solution Approach 2:
The system changes the feature point extraction parameters according to the identified scene type. For outdoor scenes, one set of parameters is used; for indoor scenes, another set; and for regular pattern scenes, a third set. This parameter adaptation enables accurate calibration for each scene type without requiring manual intervention.
3Measurement precision
If multiple calibration parameter values are integrated from different scene types, then the overall calibration accuracy improves, but the processing complexity increases
Solution Approach 1:
The patent segments the calibration process by obtaining separate calibration parameter values for different scene types (outdoor, indoor, regular pattern) and then integrating them. This segmentation approach allows the system to achieve high overall calibration accuracy by combining scene-specific calibrations while managing processing complexity through structured integration.
Solution Approach 2:
The patent creates a universal calibration system that handles multiple scene types through a single integrated process. The system performs multiple functions (outdoor calibration, indoor calibration, regular pattern calibration) within one unified framework, achieving comprehensive accuracy without requiring separate independent systems for each scene type.
Data Source
AI summary
An electronic apparatus is provided. The electronic apparatus includes a camera, a memory in which a plurality of captured images obtained through the camera and a parameter value of the camera are stored, and a processor electrically connected to the camera and the memory. The processor is configured to identify a scene type corresponding to each captured image of the plurality of captured images; extract a feature point of each captured image of the plurality of captured images based on a feature point extraction method corresponding to the identified scene type of each captured image; obtain a calibration parameter value corresponding to a feature type of each extracted feature point; obtain an integrated calibration parameter value based on one or more obtained calibration parameter values; and update the parameter value stored in the memory based on the integrated calibration parameter value.