Technical Drawing Image Parsing for Automated Dimension Association
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting technical data in technical drawings are inefficient and lack automation, particularly in converting 2D representations into numerical models for manufacturing and defect detection.
Innovation Solution
A computer-implemented method using neural networks for view-splitting, annotation-detection, and text-recognition modules to identify and associate geometric and annotation data in technical drawings, enabling the reconstruction of numerical models from images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual methods are used for detecting technical data in technical drawings, then the process is simple to implement, but the productivity is low and the process is time-consuming
Solution Approach 1:
The patent segments the technical drawing analysis into distinct modules: view identification module, annotation detection module, and text recognition module. Each module handles a specific aspect of the drawing, allowing parallel processing and improving overall productivity while keeping individual module complexity manageable
Solution Approach 2:
The patent introduces an intermediary processing layer that converts image data into structured technical data through multiple specialized modules. This intermediary structure enables automated high-speed processing while maintaining organized data flow, resolving the contradiction between automation efficiency and system complexity
2Productivity
If automated neural network methods are used for view identification and annotation detection, then the productivity increases, but the device complexity increases
Solution Approach 1:
The patent divides the automated detection system into separate neural network modules: view identification, annotation detection, and text recognition. This segmentation allows each module to be optimized independently and processed in parallel, increasing productivity while managing complexity through modular architecture
Solution Approach 2:
The patent employs universal neural network architectures that can handle multiple tasks within each module. For example, the view identification module not only identifies views but also classifies them, and the annotation detection module detects and classifies annotations simultaneously, reducing overall system complexity while maintaining high productivity
3Measurement precision
If multiple modules are applied for view identification, annotation detection, and text recognition, then the measurement precision improves, but the device complexity increases
Solution Approach 1:
The patent implements specialized modules for each detection task: view identification, annotation detection, and text recognition. Each module is optimized for its specific function, improving measurement precision through dedicated processing while managing complexity through clear module boundaries and standardized interfaces
Solution Approach 2:
The patent incorporates feedback mechanisms where the output of one module serves as input to the next, with validation and refinement at each stage. This feedback loop ensures high precision in technical data detection while maintaining organized system complexity through structured information flow
4Reliability
If graph representation is used to associate nodes with geometric and annotation data, then the reliability of data association improves, but the device complexity increases
Solution Approach 1:
The patent uses graph representation as an intermediary data structure to associate geometric and annotation data. The graph nodes and edges provide a structured medium for reliable data association, improving reliability while managing complexity through organized data relationships rather than complex processing algorithms
Data Source
Figure 1A
Figure 1A
Figure 1B
AI summary
The disclosure concerns a technical data detection method in a technical drawing image. The technical drawing comprises a view of a technical object and a technical annotation. The method comprises identifying one or more views in the technical drawing. The method comprises identifying one or more technical annotations in each view. The method comprises identifying characters in each technical annotation. The method comprises determining a graph representation of each view. The graph representation comprises nodes each corresponding to a classification of pixels in the view into a semantic class and edges each connects two nodes either if the two nodes represent neighboring pixels or if the two nodes represent pixels distant from each other below a threshold. The method comprises, for each identified view, using the graph topology and the identified characters to associate nodes corresponding to the dimension-related symbol or dimension classes to nodes corresponding to the geometry class.