3D Part Machining Feature Recognition Using Multi-View Vision Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for machining feature recognition, particularly for complex and new features, struggle with low accuracy and require extensive manual operations, leading to long process cycles and high labor costs in personalized customization.
Innovation Solution
A part machining feature recognition method based on machine vision learning recognition, involving sample training and feature recognition using 2D images from multiple angles, with image interception and mapping to 3D models to automate the recognition process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual feature definition and recognition methods are used, then the system can handle simple and standard machining features, but it fails to effectively recognize complex intersection features and new machining features, resulting in low recognition accuracy
Solution Approach 1:
The patent replaces manual feature definition and rule-based recognition with a machine learning model trained on image data. The system captures images of machining features, trains a neural network to recognize patterns, and uses the trained model to automatically identify both standard and complex features without manual intervention, thereby improving both accuracy and adaptability to new feature types
Solution Approach 2:
The patent transforms machining feature recognition from a rule-based parameter matching approach to an image-based machine learning approach. By changing the input parameters from geometric measurements to image pixels and using learned parameters from training data, the system can recognize complex intersection features and new feature types that cannot be defined by traditional geometric parameters
2Productivity
If traditional feature recognition methods are used, then the process can be automated to some extent, but it requires extensive manual operations for process preparation, resulting in long process cycles and high labor costs
Solution Approach 1:
The patent implements self-service automation where the machine learning model automatically performs feature recognition without human intervention. The system captures images, processes them through the trained model, and generates recognition results autonomously, eliminating the need for manual feature definition and classification operations, thereby significantly reducing both labor costs and process preparation time
Solution Approach 2:
The patent applies preliminary action by training the machine learning model in advance with extensive sample data before actual feature recognition. This pre-training phase enables the model to learn complex feature patterns beforehand, so that during actual production, the system can rapidly recognize features without requiring manual setup or adjustment, thus reducing process preparation cycles
Data Source
AI summary
A part machining feature recognition method based on machine vision learning recognition comprises sample training and part machining feature recognition. The sample training specifically refers to obtaining the 2D images of different 3D models of parts at different angles, marking machining feature information, constructing a 2D image sample, and then performing feature recognition training on the image recognition model using the 2D image samples; the part machining feature recognition specifically refers to taking 2D image screenshots from multiple view angles, recognizing all machining features from 2D image screenshots using the trained image recognition model, mapping the recognized machining features to the 3D model of parts with machining features to be recognized based on the view angle relationship, marking the machining features and geometric surfaces contained in each feature, and completing the automatic recognition of part machining features.