Character Recognition Using Template Images With Uniform Backgrounds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional character recognition methods face challenges with accurate recognition of characters on valuable media due to background patterns, leading to image quality deterioration and erroneous recognition, especially in banknote handling systems where speed and position fluctuations affect image processing.
Innovation Solution
A character recognition method using template images prepared with background areas filled based on intensity distribution, allowing for high-accuracy recognition by evaluating similarity without removing the background pattern, and considering various fonts and background patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If image processing such as binarizing and filtering is performed to remove the background pattern, then the background pattern can be removed from the image, but the image quality of the character portion deteriorates leading to erroneous recognition
Solution Approach 1:
The patent creates template images that are copies of the original character images with background patterns, but with the background patterns replaced by uniform colors. These template images serve as reference patterns for recognition, allowing the system to match characters against templates that have already had their background patterns removed in a controlled manner, rather than directly processing the original images and risking quality deterioration.
Solution Approach 2:
The patent performs background pattern removal in advance during the template image creation phase, not during the actual character recognition phase. By pre-processing the background removal and creating clean template images beforehand, the system avoids the need to perform destructive image processing on the original character images during recognition, thus preserving image quality while still achieving background pattern removal.
2Adaptability or versatility
If the background pattern is complex or changes at every character position, then accurate determination of background pattern effect becomes difficult, but conventional methods still attempt to remove it causing errors
Solution Approach 1:
The patent applies different uniform colors to different background regions based on their local characteristics. By analyzing the background pattern in each local region and assigning appropriate uniform colors, the system adapts to complex and varying background patterns while maintaining recognition accuracy. Each character's template is created with background colors tailored to its specific local background characteristics.
3Productivity
If banknotes are transported at high speed, then productivity increases, but the depth of field becomes small causing character blur and recognition errors
Solution Approach 1:
The patent creates template images that capture the essential character features even when captured under less than ideal conditions. By having pre-prepared templates with clear character representations, the system can match blurred or lower-quality captured images against these clear templates, compensating for the quality loss due to high-speed transport and shallow depth of field.
Data Source
Figure 1A~1C
Figure 2
Figure 3A~3C
AI summary
A character recognition method for recognizing a character printed over a background pattern on a valuable medium includes acquiring a character image captured by capturing the character printed on the valuable medium; evaluating a degree of similarity between the character image and each template image, the template image having been obtained beforehand by capturing each character that had a possibility of having been printed on the valuable medium and of which background area is evenly filled; and determining that the character corresponding to the template image showing the highest degree of similarity at the evaluating is the character included in the character image.