一种基于深度学习的多种类异物的识别方法及系统
By dynamically adjusting the illumination on the conveyor belt and using open self-generated kernel clustering, combined with multi-scale feature analysis of deep learning, the instability problem of foreign object recognition under multi-illumination environments is solved, achieving stable fusion of foreign object contours and recognition of unknown categories, thus improving the accuracy and robustness of recognition.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- GUIZHOU UNIV
- Filing Date
- 2025-11-30
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies are unstable in foreign object recognition under multi-light conditions, and have difficulty handling inconsistent foreign object contours, identification of unknown categories, and insufficient analysis of common features at multiple scales, resulting in inaccurate recognition results.
Image acquisition is performed using an image acquisition device above the conveyor belt. The light intensity is dynamically adjusted, and open self-generated kernel clustering and multi-scale feature analysis are used in conjunction with cross-frame contour indexing and deep feature extraction networks to achieve stable identification of foreign objects.
It improves the stability and accuracy of foreign object recognition, can adapt to complex lighting and changing postures, enhances the ability to process unknown categories and weak feature samples, and outputs higher precision recognition results.
Smart Images
Figure CN121837703B_ABST
Abstract
Citation Information
Patent Citations
Intelligent scanning and recognition system and method for foreign matters at vehicle bottom
CN117593592A
Railway foreign-object intrusion detection method and system based on deep learning
WO2025167410A1