Image Alignment via Correlation Filtering and Frame Trimming
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies face challenges in accurately aligning and correcting position deviations between images that use the same template, leading to misalignment of features like ruled lines and tables, which affects region specification and character recognition.
Innovation Solution
An image processing device that includes modules for receiving and splitting images, calculating correlation coefficients, excluding certain similarity values, and performing affine transformations to correct image positions, along with a trimming process to remove black pixels and frame images, ensuring accurate alignment and extraction of regions for character recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If correlation coefficient calculation is used for image alignment, then position deviation correction is achieved, but high similarity values cause misalignment of features like ruled lines and tables
Solution Approach 1:
The patent extracts and removes correlation coefficient values that exceed a predetermined threshold from the correlation coefficient map. This extraction of problematic high-similarity values prevents them from causing misalignment of features like ruled lines and tables, while retaining valid correlation coefficients for accurate position deviation correction.
Solution Approach 2:
The patent changes the parameter state of correlation coefficients by comparing each value against a threshold and selectively removing those that exceed it. This parameter-based filtering transforms the correlation coefficient map from containing all similarity values to containing only valid alignment information, resolving the contradiction between using correlation for alignment and avoiding misalignment from high similarity.
2Productivity
If region extraction is performed from images, then character recognition target is obtained, but black pixels connected to frame remain and affect recognition accuracy
Solution Approach 1:
The patent converts the harmful black pixels connected to the frame into a detectable pattern by adding a frame around the extracted region. The black pixels that would normally cause recognition errors are transformed into a visible connection pattern that can be identified and removed through frame-based detection, turning a recognition obstacle into a removable artifact.
Solution Approach 2:
The patent performs preliminary removal of black pixels connected to the frame before character recognition is executed. By pre-processing the extracted region to eliminate these problematic pixels, the system prevents recognition errors before they occur, improving both the precision and reliability of the overall character recognition process.
Data Source
AI summary
An image processing device is provided with an extracting unit that extracts a region from an image in accordance with an operation of an operator, an adding unit that adds a frame to the region, a removing unit that removes black pixels connected to the frame, and an output unit that outputs a process result of the removing unit.