Label Identifier Extraction Using Alignment Reference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing individual identification techniques struggle to universally apply to objects of different shapes and sizes, and fail when the object surface with a label has colors or patterns that are hard to distinguish from the label, requiring the entire label image for feature extraction.
Innovation Solution
An individual identifier extraction device that images a label with a random pattern and alignment reference portion, extracts a feature quantity from a specific area determined by the alignment reference portion, allowing for universal application across various object shapes and sizes, and enables feature extraction even when the object surface is difficult to distinguish from the label.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If feature quantity extraction is performed from the entire label image, then individual identification can be achieved, but the system cannot universally apply to objects of different shapes and sizes
Solution Approach 1:
The label is segmented into a specific extraction area defined by reference portions (e.g., four corner markers or edge portions). Instead of processing the entire label image, the system extracts feature quantities only from the defined area, enabling universal application across different object shapes and sizes while reducing data processing requirements
Solution Approach 2:
Reference portions (alignment marks, corner markers, or edge portions) are introduced as intermediary elements to define the extraction area. These reference portions serve as mediators between the imaging system and the label content, enabling consistent area definition regardless of the underlying object's shape or size
2Reliability
If the entire label image is used for feature extraction, then individual identifiers can be obtained, but the system fails when object surface patterns are hard to distinguish from the label
Solution Approach 1:
The system extracts only the necessary portion of the label image defined by reference portions, separating the feature extraction area from the rest of the label and object surface. This eliminates interference from object surface patterns that are hard to distinguish from the label, improving reliability by focusing only on the clearly defined label area
Solution Approach 2:
The reference portions (corner markers, edge portions) are designed with distinct local characteristics that differ from the random pattern area. This local quality difference enables reliable detection of the extraction area boundaries, allowing the system to accurately define the feature extraction region regardless of object surface characteristics
Data Source
AI summary
An individual identifier extraction device includes: an imaging part configured to obtain an image by imaging a label which is attached to an object and includes a random pattern and an alignment reference portion; and an extraction part configured to acquire, from the image obtained by imaging, an image of an area determined with reference to the alignment reference portion on the label, and extract a feature quantity of the random pattern as an individual identifier of the object from the image of the area.