Component Recognition via Edge Extraction and Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing component recognizing methods in factory automation systems are inefficient due to sensitivity to illumination changes and complexity, making real-time processing difficult and requiring new algorithms for each new product, which hinders the application of component recognition technology in varying production environments.
Innovation Solution
A component recognizing apparatus and method that uses edge detection techniques like second derivative calculation and local adaptive binarization to extract feature vectors from component images, which are then input into an artificial neural network for category recognition and posture angle estimation, enhancing robustness against illumination variations and enabling recognition of various components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If model-based matching method or rule-based method is used for component recognition, then component recognition can be performed, but the system requires new algorithms for each new product and is sensitive to illumination changes
Solution Approach 1:
The patent transforms the component recognition problem from pixel-based matching to shape-based recognition by changing the parameters used for comparison. It extracts contour lines and converts them to shape files, then uses shape matching algorithms that compare geometric parameters rather than pixel values, making the system invariant to illumination changes and adaptable to new products.
Solution Approach 2:
The patent replaces the traditional model-based matching mechanism with a shape-based recognition mechanism. Instead of comparing pixel intensities or using complex rule-based systems, it substitutes a geometric shape matching approach that uses contour extraction and shape file comparison, simplifying the system and improving adaptability.
2Reliability
If complex matching algorithms are used to improve recognition accuracy, then component recognition rate improves, but real-time processing becomes difficult
Solution Approach 1:
The patent extracts only the essential features for recognition by isolating contour lines and converting them to shape files, discarding redundant pixel information. This extraction process reduces the data volume significantly while retaining the key geometric characteristics needed for accurate recognition, enabling real-time processing.
Solution Approach 2:
The patent segments the component image processing into distinct stages: contour line extraction, shape file conversion, and shape matching. This segmentation allows each stage to be optimized independently, with the contour extraction using efficient edge detection algorithms and the matching using simplified geometric comparisons, improving overall processing speed.
3Measurement precision
If traditional edge detection methods are used, then component edges can be detected, but the detection is sensitive to illumination variations
Solution Approach 1:
The patent inverts the traditional approach by not directly detecting edges from pixel intensity gradients, but instead first converting the image to a shape representation through contour extraction. This inversion makes the detection process invariant to illumination changes because it operates on geometric boundaries rather than intensity values.
Data Source
AI summary
Disclosed are a component recognizing apparatus and a component recognizing method. The component recognizing apparatus includes: an image preprocessing unit configured to extract component edges from an input component image by using a plurality of edge detecting techniques, and detect a component region by using the extracted component edges; a feature extracting unit configured to extract a component feature from the detected component region, and create a feature vector by using the component feature; and a component recognizing unit configured to input the created feature vector to an artificial neural network which has learned in advance to recognize a component category through a plurality of component image samples, and recognize the component category according to a result.