Image Stitching Quality Evaluation and Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing apparatuses do not evaluate the quality of stitched images, resulting in unnatural-looking outputs.
Innovation Solution
An image processing apparatus that includes an image synthesizer for correcting input images using first correction data, a determiner to assess the appropriateness of stitching using a trained model, a second correction data generator to generate second correction data, and an image updater to update the synthesized image based on the assessment, ensuring natural-looking stitched images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional stitching processing is executed without quality evaluation, then the processing speed is maintained, but the image quality becomes unnatural
Solution Approach 1:
The patent implements feedback by using a trained model to evaluate the stitching quality of generated images and feed this evaluation back to the image synthesizer. The determiner assesses whether images are appropriately stitched together, and this information is used to iteratively refine the stitching process, thereby improving image quality while managing processing complexity through automated evaluation loops.
Solution Approach 2:
The system performs self-service by automatically evaluating its own output quality using the trained model and self-correcting the stitching process based on the evaluation results. The image processing apparatus autonomously identifies quality issues and adjusts parameters without external intervention, enabling continuous improvement of stitching quality while maintaining operational efficiency.
2Manufacturing precision
If a trained model is used to evaluate stitching quality, then the image quality improves, but the processing time increases
Solution Approach 1:
The patent applies preliminary action by pre-training the model offline using training images and their corresponding ground truth labels. This training phase is performed in advance, allowing the model to be ready for rapid evaluation during actual stitching operations. The pre-trained model can then quickly assess stitching quality without requiring extensive processing time during the main workflow.
Solution Approach 2:
The system uses partial action by selectively applying the trained model evaluation only to critical stitching cases or using a simplified evaluation metric for routine cases. This approach balances the need for quality assurance with processing time constraints, applying comprehensive model evaluation only when necessary rather than to every single stitching operation.
3Manufacturing precision
If iterative correction using second trained model is implemented, then the stitching accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent segments the stitching process into distinct phases: an initial stitching phase using first correction data, followed by an evaluation phase using the first trained model, and then a refinement phase using the second trained model. This segmentation allows each component to be optimized independently, managing overall system complexity while achieving high stitching accuracy through coordinated multi-stage processing.
Solution Approach 2:
The system employs parameter changes by adjusting correction parameters based on evaluation results. The second trained model generates updated correction data that modifies stitching parameters iteratively. This dynamic parameter adjustment enables the system to adapt to different image characteristics and achieve higher accuracy without requiring a fundamentally more complex system architecture.
Data Source
AI summary
An image processing apparatus includes an image synthesizer configured to correct input images based on first-correction data to generate corrected images, and to generate a synthesized image by stitching the corrected images together, a determiner configured to determine whether the corrected images are appropriately stitched together in the synthesized image by using a first-trained model learned whether images are appropriately stitched together, a second-correction-data generator configured to generate second-correction data by supplying the input images to a second-trained model learned relationships between correction data used to correct source images to generate corrected images appropriately stitched together and the source images, and an image updater configured to output the synthesized image when a determination result of the determiner is affirmative and output an updated synthesized image generated by causing the image synthesizer to update the synthesized image based on the second-correction data when the determination result of the determiner is negative.