Printed Image Inspection Alignment With Conditional Nonlinear Warping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing inspection systems for printed materials face inaccuracies in position adjustment between images due to local distortions, leading to erroneous inspections, particularly when non-linear transformations are applied without proper consideration of image content differences.
Innovation Solution
An inspection apparatus that performs position adjustment by linear transformation initially, and if a certain condition is met, applies non-linear transformation, such as Free-Form Deformations (FFD), based on the presence of printed objects or data type in the image, to prevent erroneous inspections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If position adjustment by non-linear transformation is performed to handle local distortion, then position adjustment accuracy is improved, but the risk of erroneous inspection increases when objects differ between images
Solution Approach 1:
The patent segments the position adjustment process into two distinct stages: first performing linear transformation to achieve coarse alignment, then conditionally applying non-linear transformation for fine adjustment. This segmentation allows the system to benefit from the accuracy of non-linear transformation while avoiding its reliability issues by limiting its application to cases where local distortion is present and objects match between images.
Solution Approach 2:
The patent applies linear transformation as a preliminary action before attempting non-linear transformation. This preliminary alignment ensures that the images are roughly matched in position and orientation first, which provides a stable foundation for the subsequent conditional non-linear adjustment and reduces the risk of erroneous matching.
2Device complexity
If position adjustment by linear transformation is used for simplicity, then device complexity is reduced, but position adjustment accuracy deteriorates due to inability to handle local distortion
Solution Approach 1:
The patent implements a dynamic position adjustment system that adapts its complexity based on the characteristics of the input images. The system automatically selects between linear and non-linear transformation methods depending on whether local distortion is detected, allowing the complexity of the adjustment process to vary dynamically rather than being fixed, thus balancing simplicity and accuracy.
Solution Approach 2:
The patent changes the transformation parameter (from linear to non-linear) based on the detected needs of the specific inspection case. By monitoring image characteristics and switching between different transformation models, the system optimizes the balance between computational complexity and adjustment accuracy for each particular situation.
Data Source
AI summary
The occurrence of an erroneous inspection is suppressed by preventing a failure of position adjustment by non-linear transformation. In position adjustment between an inspection image obtained by reading a printed material and a reference image, after performing position adjustment by linear transformation, only in a case where a certain condition is satisfied, position adjustment by non-linear transformation is further performed for the inspection image for which the position adjustment by linear transformation has been performed.