Signal Processing Device for Accurate Pattern Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information reading apparatuses face challenges in accurately correcting non-uniformity in sensing signal levels during the reading period due to variations in environmental conditions and sensor aging, leading to inadequate signal correction and reduced accuracy in pattern recognition.
Innovation Solution
A signal processing device that includes a pattern presence/absence evaluation unit, a correction data updating unit, and a correction data subtraction unit, which calculates and updates correction data based on the degree of signal inclusion and applies weighted sums to correct sensing data in real-time, effectively reducing signal variations and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If offset adjustment is performed only in the interval between paper currency passages, then the device complexity is reduced, but the measurement precision deteriorates because non-uniformity cannot be adequately corrected when the sensing signal level varies during reading
Solution Approach 1:
The patent implements continuous offset adjustment during the entire reading period by updating correction values for each sensing element at every reading position. This continuous correction process ensures that non-uniformity in sensing signal levels is adequately compensated throughout the reading interval, resolving the contradiction between adjustment frequency and measurement precision.
Solution Approach 2:
The patent employs feedback mechanisms where the sensing signal level is continuously monitored and used to update correction values dynamically. The correction values are adjusted based on the actual sensing signal levels detected at each position, creating a closed-loop system that maintains measurement precision without requiring excessive adjustment frequency.
2Measurement precision
If continuous correction is applied during the reading interval, then the measurement precision is improved, but the processing time increases
Solution Approach 1:
The patent performs preliminary correction by pre-calculating and storing correction values for each sensing element based on their individual characteristics. These pre-computed correction values are then applied during the reading process, eliminating the need for complex real-time calculations and reducing processing time while maintaining correction accuracy.
Solution Approach 2:
The patent divides the correction process into independent segments for each sensing element. By treating each sensing element's correction separately and using individual correction values, the system can efficiently process corrections without requiring comprehensive real-time analysis of all sensing elements, thereby reducing overall processing time.
3Measurement precision
If the offset adjustment value is updated frequently, then the measurement precision is improved, but the loss of information increases due to more frequent calculations
Solution Approach 1:
The patent applies local quality by generating individual correction values for each sensing element based on their specific characteristics and performance. This localized correction approach ensures that each sensing element is corrected according to its own non-uniformity patterns, improving overall measurement precision while avoiding the need for extensive global calculations that would increase information loss.
Data Source
AI summary
A pattern presence/absence evaluation value generation unit (41) determines the presence or absence of a pattern by comparing sensing data (d3) and correction data (d44). Operating according to a pattern presence/absence evaluation value (d41) obtained from the pattern presence/absence evaluation value generation unit (41), a correction data updating unit (42) outputs correction data (d42) updated by weighted addition of the sensing data (d3) and the correction data (d44), increasing the weighting of the correction data (d44) when an image pattern is present and increasing the weighting of the sensing data (d3) when the image pattern is absent. A correction data subtraction unit (43) subtracts the correction data (d42) from the sensing data (d3) and outputs output sensing data (d4). It consequently becomes possible to reduce the effect of sensing signal level variations of the sensing signal generated within the interval of reading of the object being sensed, and read the pattern accurately.