Shingle Defect Detection via Multi-Stage Image Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The shingle industry faces challenges in accurately differentiating between real defects and false defects using machine vision, and in training sawyers or programming computers to classify shingles according to complex CSA standards, leading to inefficiencies and suboptimal shingle production.
Innovation Solution
A post-sawing quality control system that includes a machine vision system connected to a computer, an image tracking system, and a carriage tracking system, which picks sawed shingles, inspects them for defects, and maintains a database of images to differentiate between real and false defects, ultimately improving shingle classification and packaging efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If machine vision system is used to detect defects, then productivity is improved, but measurement precision deteriorates due to inability to differentiate real defects from false defects
Solution Approach 1:
The system performs preliminary actions by capturing images of the wood block at multiple stages (before sawing, during sawing, and after sawing) and storing them in a database. This preliminary capture and storage of image data enables subsequent comparison and analysis to distinguish real defects from false defects, thereby improving measurement precision while maintaining productivity
Solution Approach 2:
The system implements feedback by comparing newly captured defect images with previously stored images in the database. The computer analyzes patterns across multiple images to determine whether detected defects are real or false, providing feedback that improves defect detection accuracy without reducing production speed
2Ease of operation
If computer is programmed to classify shingles according to CSA standards, then ease of operation is improved, but device complexity increases due to difficulty of programming complex classification criteria
Solution Approach 1:
The system creates copies of the complex classification rules by capturing and storing visual examples of defects and quality criteria in an image database. Instead of programming complex logical rules, the computer learns from visual copies of actual defects and quality standards, simplifying the programming task while maintaining accurate classification according to CSA standards
Solution Approach 2:
The system replaces the mechanical approach of programming explicit classification rules with an optical/image-based approach. The computer vision system analyzes visual patterns directly from images, substituting complex logical programming with image recognition algorithms that naturally handle the complexity of CSA classification criteria
3Manufacturing precision
If all defects are removed from shingles, then manufacturing precision is improved, but loss of substance increases due to removal of potentially usable wood
Solution Approach 1:
The system performs preliminary analysis by capturing and storing images of the wood block at multiple stages before final processing. This allows the computer to predict defect locations and assess their impact on shingle usability in advance, enabling selective removal of only those defects that truly compromise quality, thereby reducing unnecessary wood waste while maintaining manufacturing precision
Solution Approach 2:
The system applies local quality assessment by analyzing defects in specific locations and contexts rather than applying uniform removal criteria. The computer evaluates each defect's impact on shingle quality based on its location, size, and type, allowing preservation of wood in areas where defects do not compromise overall quality, thus reducing waste while maintaining precision
Data Source
AI summary
In a first aspect, there is provided a system for picking sawed shingles against a saw in movement. Further, F in a method for maintaining a database of images of shingle defects, wherein a front-face image and a backside image of a shingle are associated with each other in that database. To increase shingle quality, each shingle is inspected on five faces thereof, to detect surface defects and core defects. Comparison is made of images of the front-face to images of wood defects in the database. When the image of the front-face of a shingle matches an image of an acceptable defect, and that front-face image is tagged as “predisposed to backside defect”, the shingle is edged to remove the acceptable defect. In the shingle manufacturing process, each of these backside images is considered to be a mirror image of a next shingle to be sawed.