Print Medium Identification Using Dual Reflectance Discriminators
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to accurately identify the type of print medium with subtle differences due to averaging out of distinct physical properties when using a single machine learning model.
Innovation Solution
A method and system utilizing two pre-trained discriminators, each processing different types of physical property information (spectral reflectance and reflectance distribution) to identify print media, leveraging vector neural networks for precise classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single machine learning model is used to process multiple types of physical property information, then the system complexity is reduced, but the identification accuracy deteriorates due to averaging out of distinct physical properties
Solution Approach 1:
The patent divides the identification system into multiple independent discriminators, each dedicated to processing specific types of physical property information (e.g., spectral reflectance, reflectance distribution, basis weight). This segmentation prevents the averaging effect that occurs in single-model approaches, allowing each discriminator to specialize in extracting features from its designated input type, thereby maintaining high identification accuracy while managing system complexity through modular architecture
2Measurement precision
If multiple pre-trained discriminators are used to process different physical property information, then the identification accuracy improves, but the device complexity increases
Solution Approach 1:
The patent merges the outputs of multiple independent discriminators through a synthesis unit that combines discrimination information from different physical property measurements. This merging strategy allows the system to leverage the specialized capabilities of each discriminator while integrating their results to achieve superior overall identification accuracy, balancing the increased device complexity with enhanced performance
3Measurement precision
If multiple types of physical property information are processed separately, then the identification of subtle differences improves, but the loss of information increases due to separate processing paths
Solution Approach 1:
The patent implements feedback mechanisms where the synthesis unit receives discrimination information from multiple discriminators and integrates it to form a comprehensive identification result. This feedback loop ensures that information from different physical property measurements is preserved and combined effectively, preventing information loss while maintaining the ability to detect subtle differences through specialized processing paths
Data Source
AI summary
The print medium identification method includes (a) a step of obtaining first physical property information about a print medium, (b) a step of obtaining second physical property information which is different from the first physical property information about the print medium, (c) a step of obtaining first discrimination information for discriminating a type of print medium by inputting the first physical property information to a first discriminator, (d) a step of obtaining second discrimination information for discriminating the type of print medium by inputting second physical property information to a second discriminator, and (e) a step of identifying the type of print medium by using the first discrimination information and the second discrimination information.