Cigarette End-Face Inspection Using Radial Contour Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In the tobacco industry, particularly in cigarette production, automatic recognition of products with circular end faces in images is challenging due to varying positions and the need for new reference patterns for each formation, making the process time-consuming and costly.
Innovation Solution
A method using electro-optical testing devices to evaluate images along multiple evaluation lines, determining edge points of the product contour, and calculating the center of a circle through these points to accurately position and recognize products without relying on reference images, allowing for automated detection and defect evaluation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If reference patterns are stored for each different cigarette formation, then product recognition accuracy is improved, but device complexity and time consumption increase
Solution Approach 1:
The system automatically determines product positions and contours by evaluating image information along multiple evaluation lines without requiring pre-stored reference patterns. The method self-calibrates by calculating the center of the circle through edge points directly from the captured image, eliminating the need for external reference data storage and management.
Solution Approach 2:
The invention extracts only the necessary information (edge points on evaluation lines) from the image to determine product positions and contours. By focusing solely on contour detection through evaluation lines rather than storing complete reference patterns, the system reduces complexity while maintaining recognition accuracy.
2Measurement precision
If reference patterns are stored for each different cigarette formation, then product recognition accuracy is improved, but productivity decreases
Solution Approach 1:
The system performs automatic self-calibration by evaluating image information along multiple evaluation lines and calculating product positions directly from edge points. This eliminates the time-consuming process of selecting and loading appropriate reference patterns for different formations, enabling continuous high-speed operation without productivity losses.
Solution Approach 2:
The method pre-defines multiple evaluation lines radiating from the expected product center before image capture. This preliminary setup allows immediate processing of any product formation without requiring post-capture reference pattern selection, maintaining high productivity while ensuring accurate recognition.
3Measurement precision
If multiple evaluation lines are used to determine product position, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The image evaluation is segmented into multiple independent evaluation lines radiating from the expected product center. Each evaluation line independently identifies edge points and contributes to determining the final product position and contour. This segmentation simplifies the overall complexity by breaking down the two-dimensional recognition problem into multiple one-dimensional line evaluations.
Solution Approach 2:
The invention transitions from two-dimensional pattern matching to a dimensional approach by using multiple radial evaluation lines at different angles. This multi-dimensional evaluation method improves position accuracy by cross-validating edge point detections across different angular perspectives while keeping each individual line evaluation computationally simple.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
The invention relates to a method and an apparatus for checking products (10, 48) in the tobacco industry with a respectively essentially circular end contour (40, 53), particularly cigarettes, in which an electro-optical checking element (23), preferably a camera, records an electronic image of the end (24, 55) of at least one product (10, 48), wherein the recorded image (47, 50) is evaluated for the purpose of identifying the product (10, 48) in the image (47, 50). The invention is characterized in that the evaluation of image information in the recorded image (47, 50) along at least three evaluation lines (31), which each extend from an inner image position (32), at which the end (24, 55) of the product (10, 48) is expected in the image (47, 50), to an outer image position (33), at which the surroundings of the end (24, 55) of the product (10, 48) are expected in the image (47, 50), ascertains boundary points for the end contour (40, 53) of the product (10, 48) which are situated on the evaluation lines (31), with the centre (43) of a circle that extends through the boundary points being calculated from at least three ascertained boundary points.