A multi-modal structured parsing method and system for energy industry drawings
By employing a multimodal structured parsing method, the problem of drawing quality was solved, enabling the complete extraction and semantic understanding of drawing information, and supporting the digital management and value release of drawing assets.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CNOOC ENERGY TECHNOLOGY & SERVICES LTD
- Filing Date
- 2026-04-17
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies struggle to effectively address the issues of inconsistent quality, complex and diverse information formats, and strong implicit semantic relationships in scanned drawings. This makes it difficult to digitize and implement drawing information, hindering the formation of a structured data system and limiting the reuse and value release of drawing knowledge.
A multimodal structured parsing method is adopted, including image preprocessing, primitive detection and vectorization, arbitrary-direction text detection and multimodal semantic fusion. Combining computer vision, optical character recognition and natural language processing technologies, the association between text and graphics is established, and a topological structure representation is constructed and stored in the database.
It enables complete extraction and semantic understanding of drawing information, supports equipment management, process traceability and intelligent retrieval, and enhances the digital value of drawing assets.
Smart Images
Figure CN122416482A_ABST