Scanned Image Distortion Correction Using Circumscribed Rectangles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image distortion correction methods face challenges in accurately correcting distortion in scanned images when characters have aspect ratios not equal to 1 and when characters are closely located in the scanning directions, leading to improper formation of minimal circumscribing rectangles and character lines.
Innovation Solution
An image distortion correction apparatus and method that uses a scanner and image data processor to form circumscribed rectangles and extract character lines, adjusting the rectangle lengths in the main scanning direction to correct distortion, even when characters are closely located, by determining correction ratios and selecting optimal character lines based on predetermined criteria.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If minimal circumscribing rectangles are formed based on character aspect ratio for distortion correction, then distortion correction accuracy is improved when aspect ratio is 1, but correction accuracy deteriorates when aspect ratio is larger or smaller than 1
Solution Approach 1:
The patent changes the basis for distortion correction from character aspect ratio to the ratio of circumscribing rectangle lengths in main scanning and sub-scanning directions. This parameter change allows the method to work effectively regardless of character aspect ratio, resolving the contradiction between correction accuracy and adaptability to various aspect ratios.
2Manufacturing precision
If minimal circumscribing rectangles are formed for each character, then distortion correction can be performed, but proper rectangle formation becomes difficult when characters are located closely in the sub-scanning direction
Solution Approach 1:
The patent merges adjacent circumscribing rectangles in the sub-scanning direction when characters are closely located, forming a combined rectangle that encompasses multiple characters. This merging approach allows distortion correction to proceed even when individual character rectangles cannot be properly formed due to close spacing.
3Measurement precision
If minimal circumscribing rectangles are used to determine character lines, then character line extraction can be performed, but proper character line determination becomes difficult when characters are located closely in the main scanning direction
Solution Approach 1:
The patent addresses character line determination by considering the spatial distribution of circumscribing rectangles in both main scanning and sub-scanning directions. When characters are closely located in the main scanning direction, the method uses the vertical positioning information from the sub-scanning direction to properly distinguish and determine multiple character lines, adding a dimensional perspective to the analysis.
Data Source
AI summary
An apparatus, system, method, and computer program product is disclosed, each capable of correcting distortion in a scanned image, using at least a character line extracted from the scanned image. The character line is extracted based on a circumscribed rectangle, representing the vertical component of the character. The distortion in the scanned image is corrected based on the length of the circumscribed rectangle in the main scanning direction.