Inspection Record Recognition Using Auto-Detected Region Coordinates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for managing coordinate values of information recognition target regions in inspection records are cumbersome and require manual updates when reference images change, leading to potential errors and inefficiencies in anchor-based matching.
Innovation Solution
A computer vision-based method that extracts box regions, detects target region coordinates, and uses perspective transformation based on vertically crossing points of outermost lines to align scales, eliminating the need for manual coordinate input and anchor-based affine conversion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If anchor-based affine conversion is used to match scanned image with reference image, then information recognition can be performed, but coordinate value management becomes cumbersome and requires manual updates when reference image changes
Solution Approach 1:
The patent replaces the manual mechanical process of updating coordinate values in configuration files with an automated computer vision system. The system automatically detects target regions by analyzing image content and extracting coordinates through algorithms, eliminating the need for manual coordinate management and configuration file updates when reference images change.
Solution Approach 2:
The system enables self-service by automatically adapting to reference image changes without human intervention. When the reference image is updated, the system automatically re-detects target regions and updates coordinates through image processing algorithms, allowing the system to serve itself rather than requiring manual reconfiguration.
2Productivity
If manual coordinate input and configuration file updates are required, then coordinate precision can be maintained, but productivity decreases due to manual effort and potential errors
Solution Approach 1:
The patent replaces manual coordinate input and configuration file editing with automated computer vision algorithms. The system automatically detects target regions, extracts coordinates, and manages the recognition process without human intervention, thereby improving productivity while maintaining reliability through algorithmic precision.
Solution Approach 2:
The system incorporates feedback mechanisms where the detected target region coordinates are automatically used to adjust and refine the information recognition process. This feedback loop ensures continuous optimization of recognition accuracy while eliminating manual coordination steps, thereby improving both productivity and reliability.
3Adaptability or versatility
If anchor-based matching method is used, then scale conversion can be performed, but the method cannot cope with image size and angle changes in scanning situations
Solution Approach 1:
The patent implements dynamic adaptation by using computer vision algorithms that can automatically adjust to varying image sizes and angles. The system dynamically detects target regions and calculates appropriate coordinates based on the actual scanned image characteristics, enabling it to cope with different scanning conditions while maintaining precision through adaptive algorithms.
Solution Approach 2:
The system handles parameter changes by automatically adjusting detection and coordinate extraction parameters based on the scanned image's size and angle. The computer vision algorithms modify their behavior according to the input image characteristics, allowing the system to adapt to different scanning conditions while maintaining accurate scale conversion and coordinate precision.
Data Source
AI summary
A computer vision-based inspection record recognition method includes: extracting a box region from an inspection record reference image; detecting, within the box region, one or more coordinates of an information recognition target region; converting a scale of a target inspection record to match a scale of the inspection record reference image; and recognizing, based on the one or more coordinates of the information recognition target region of the inspection record reference image, information corresponding to same coordinates within the target inspection record that has a same scale as the inspection record reference image.