Conveyance Line Print Error Detection Using Vision and Deep Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Line-based printing systems face challenges in detecting errors, which are costly and difficult to identify, especially with traditional manual inspection methods, and existing automated approaches are resource-intensive, unreliable, and difficult to generalize across different product types.
Innovation Solution
A hybrid approach using computer vision and deep learning-based defect recognition models for automated defect detection and localization, eliminating the need for predefined thresholds or reference images, and allowing adaptability across various manufacturing applications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual inspection methods are used for error detection in line-based printing, then operational simplicity is maintained, but detection reliability and precision deteriorate due to human error and difficulty in identifying errors
Solution Approach 1:
The patent replaces manual visual inspection with an automated optical inspection system that captures images of printed products and uses image processing algorithms to detect printing errors. This substitution eliminates human error while maintaining operational simplicity through automated analysis of captured images.
Solution Approach 2:
The system creates digital copies (images) of the printed products and analyzes these copies to detect errors. By working with image copies rather than directly manipulating physical products, the system achieves high detection reliability while keeping the inspection process non-intrusive and simple to operate.
2Productivity
If traditional automated inspection approaches are used, then productivity is improved through automation, but resource intensity and device complexity increase making the system unreliable and difficult to generalize
Solution Approach 1:
The patent develops a universal inspection system that can detect multiple types of printing errors (missing lines, extra lines, misalignment, color errors) across different product types using the same core image processing algorithms. This multi-functionality achieves high productivity while avoiding the need for complex specialized systems for each error type.
Solution Approach 2:
The system achieves adaptability across different products by adjusting processing parameters such as threshold values, color space conversions, and analysis algorithms rather than changing the fundamental system architecture. This allows high productivity through automation while keeping device complexity manageable through parameter-based adaptation.
3Measurement precision
If existing automated defect detection systems are deployed, then measurement precision is improved, but adaptability deteriorates due to difficulty in generalizing across different product types
Solution Approach 1:
The patent implements a dynamic inspection system that adapts its analysis parameters and algorithms based on the specific product type and printing requirements being inspected. The system can dynamically adjust threshold values, color analysis methods, and pattern recognition parameters to maintain high detection precision across diverse product types without requiring complete system reconfiguration.
Data Source
AI summary
Systems and techniques may generally be used for error detection in an image. An example technique may include identifying a set of training images, the set of training images representing respective products on a conveyance line, the respective products being printed on before coming off the conveyance line, and labeling the set of training images with an indication of whether a product in an image in the set of training images includes a printing defect or does not including a printing defect. The example technique may include training a machine learning model using the set of training images, and outputting the machine learning model to identify printing errors on the conveyance line.