Image Correction via Affine Transformation Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
When capturing an image of an object like an analog meter with a handheld imaging device, the inclination of the imaging direction can cause positional, size, and shape deviations between the target and master image data, leading to inaccurate correction of the target image data based on the master image data.
Innovation Solution
An image processing device that generates a transformed image group through affine transformations of the target image data, calculates correlation values with the master image data, selects the transformed image data with the maximum correlation, and corrects the target image data based on positional deviations between the object in the master and target images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the imaging device is held in hand to capture images, then the ease of operation is improved, but the imaging direction becomes inclined causing positional deviation, size deviation, and shape deviation between target and master image data
Solution Approach 1:
The patent applies parameter changes by performing affine transformation on the master image data to generate multiple transformed images with different parameters (rotation angles, scaling factors, and translation vectors). This allows the system to adapt the master image to match various inclined imaging conditions, thereby maintaining correction accuracy even when the imaging device is held in hand.
Solution Approach 2:
The patent implements dynamics by creating a dynamic set of transformed master images that can adapt to different imaging conditions. The system calculates correlation values between the target image and multiple transformed master images, selecting the best match based on real-time conditions, thus enabling accurate correction despite variations in imaging angle and distance.
2Measurement precision
If affine transformation is performed to generate multiple transformed images, then the correction accuracy is improved, but the device complexity and processing time increase
Solution Approach 1:
The patent applies segmentation by dividing the correction process into distinct stages: generating transformed images with specific parameters, calculating correlation values, selecting the best match, and performing final correction. This segmented approach manages complexity by organizing the processing into manageable steps while maintaining high correction accuracy.
Solution Approach 2:
The patent uses partial action by generating a limited set of transformed images with specific parameter ranges rather than exhaustively processing all possible transformations. This approach achieves sufficient correction accuracy for practical imaging conditions while avoiding excessive processing complexity and time consumption.
3Measurement precision
If multiple transformed images are generated and correlated with master image data, then the correction accuracy is improved, but the loss of time increases due to additional processing steps
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing multiple transformed versions of the master image data with anticipated parameter variations. This preparation work is done in advance, allowing the system to quickly compare the target image against pre-prepared transformed images during actual correction, thereby reducing real-time processing time while maintaining high accuracy.
Data Source
AI summary
In the present invention, master image data is used to improve precision for correcting target image data. A generation unit (11) performs an affine transformation on the target image data, thereby generating a transformation image group including a plurality of transformation image data; a calculation unit (12) calculates a correlation value between transformation image data included in the transformation image group, and preregistered master image data; a selection unit (13) selects the transformation image data having the greatest correlation value from among the transformation image data included in the transformation image group; and a correction unit (14) corrects the selected transformation image data on the basis of the positional displacement between an object shown in the master image data and the object shown in the selected transformation image data.