Banknote Serial Number Recognition Using Dual Learning Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing banknote inspection devices face challenges in accurately recognizing serial numbers, especially when characters have holes or varying orientations, which can lead to incorrect identification and denomination discrimination.
Innovation Solution
A banknote inspection device employing a storage unit with two learning models, one trained on images with holes and another on images without holes, and a recognition unit that uses these models to accurately identify serial numbers by correcting orientation and applying binarization processing to enhance character recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single learning model is used for character recognition, then the device complexity is reduced, but the recognition accuracy deteriorates when characters have holes or varying orientations
Solution Approach 1:
The patent divides the character recognition task into two separate learning models: one specialized for recognizing characters with holes (such as '0', '6', '8', '9') and another for characters without holes (such as '1', '2', '3', '4', '5', '7'). This segmentation allows each model to be optimized for its specific character type, improving overall recognition accuracy while keeping individual model complexity manageable
Solution Approach 2:
The patent applies different recognition approaches locally based on character properties. By detecting whether a character has holes, the system selectively applies the appropriate learning model (first learning model for holed characters, second learning model for non-holed characters), ensuring that each character is processed with the most suitable recognition method for its specific features
2Measurement precision
If orientation correction and binarization processing are applied, then character recognition accuracy is improved, but the processing time increases
Solution Approach 1:
The patent performs orientation correction and binarization processing as preliminary steps before character recognition. By pre-processing the character images to correct orientation and convert to binary format beforehand, the actual recognition process becomes faster and more accurate, as the learning models receive pre-optimized input data that requires less computational effort during recognition
Data Source
AI summary
In a banknote inspection device, a storage unit stores a first learning model generated using an image of a character with a hole as training data, and a second learning model generated using an image of a character without a hole as training data, and a recognition unit recognizes a serial number character that is a character forming a serial number of a banknote by using the first learning model when a character image, which is as image of the serial number character, has a hole, and recognize the serial number character by using the second learning model when the character image does not have a hole.