PCB Character Recognition Using Fragmented Search Regions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The production of packaged modules, such as power amplifiers, is hindered by defects caused by impurities, scratches, and damage during fabrication and packaging, leading to monetary and resource losses, as well as reduced yield due to the need to identify and remove defective modules from production lines.
Innovation Solution
A circuit board processing system that uses a character recognition system to analyze digital images of characters on panels, applying local and global threshold values, along with contrast equalization and fragmented search regions, to classify characters effectively, even when damaged, thereby reducing processing time and increasing accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image analysis methods are used for character recognition on damaged circuit boards, then the system can process characters, but the processing time is excessive and accuracy is reduced due to molding flashing and damage
Solution Approach 1:
The patent divides the character recognition process into multiple processing stages: initial full-image analysis, followed by fragmented search region analysis. The character search space is segmented into multiple regions, and only promising regions are analyzed in detail. This segmentation allows the system to quickly eliminate non-matching characters while performing thorough analysis only where needed, thereby maintaining high accuracy while reducing overall processing time.
Solution Approach 2:
The patent applies different processing strategies to different regions of the image. Full image analysis is performed initially to establish baseline characteristics, then fragmented search regions are identified and analyzed with focused attention. Local threshold values are applied to specific regions rather than uniformly across the entire image, allowing adaptive processing that maintains accuracy in damaged areas while reducing computation in undamaged regions.
2Reliability
If the system performs thorough image analysis to accurately classify damaged characters, then recognition accuracy improves, but processing time increases significantly
Solution Approach 1:
The patent performs preliminary full-image analysis before detailed character classification. This initial pass establishes baseline characteristics, identifies potential damage regions, and pre-processes the image data. By performing this preliminary action, the system prepares the data structure and identifies critical regions beforehand, so that subsequent detailed analysis can be focused and efficient, maintaining reliability while improving throughput.
Solution Approach 2:
The patent implements a two-stage approach where obviously damaged or non-matching characters are quickly identified and skipped in the detailed analysis phase. The fragmented search region technique allows the system to rush through portions of the image that don't require thorough analysis, while concentrating computational resources on critical regions. This skipping strategy maintains classification reliability for important characters while dramatically improving overall processing speed.
3Manufacturing precision
If local threshold values are applied to each locality based on pixel information, then binarization accuracy improves for damaged characters, but computational complexity increases
Solution Approach 1:
The patent segments the image into multiple localities or regions before applying local threshold values. Rather than computing thresholds for every pixel across the entire image, the system divides the image into manageable regions and applies thresholding locally to each segment. This segmentation reduces the overall computational complexity while maintaining binarization precision, as each local region can be processed independently with simpler calculations.
Solution Approach 2:
The patent applies local threshold values specifically to regions where damage or molding flashing is detected, rather than uniformly across the entire image. The thresholding strategy adapts to local image characteristics, using more sophisticated local thresholding only where needed to maintain precision, while using simpler methods in undamaged regions. This local quality approach maintains binarization precision for critical areas while reducing overall computational complexity.
Data Source
AI summary
A circuit board processing system configured to process a panel, the system comprising a character recognition system configured to obtain a digital image of a character imprinted on a surface of the panel. The character recognition system is further configured to apply a fragmented search region to the digital image to obtain less than a whole of the character and apply image analysis to the less than a whole of the character; and to classify the character based on the image analysis.