Automobile seat skeleton processing defect detection method based on machine vision

CN121329956BActive Publication Date: 2026-07-03重庆飞驰汽车系统有限公司
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
重庆飞驰汽车系统有限公司
Filing Date
2025-11-12
Publication Date
2026-07-03

AI Technical Summary

Technical Problem

Existing technologies struggle to accurately identify minute defects in automotive seat frame manufacturing, such as microcracks, weld detachment, and hole deviations. They are particularly prone to low accuracy under complex surface conditions and lack full-process closed-loop control.

Method used

By employing multi-angle synchronous image acquisition and combining it with deep neural networks to extract multi-scale structural features, and using an attention mechanism to weighted focus on high-risk areas, we can perform small-sample abnormal pattern recognition and joint distribution model judgment to achieve sub-pixel-level coordinate labeling and source tracing analysis.

Benefits of technology

It enables accurate identification of the type of defects in the processing of automotive seat frames, improves the identification capability and system robustness, and provides visualized quality control support.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121329956B_ABST
    Figure CN121329956B_ABST
Patent Text Reader

Abstract

The application discloses a machine vision-based automobile seat skeleton processing defect detection method and particularly relates to the technical field of defect detection; aiming at complex defects such as virtual welding, micro-cracks, hole position deviation and collapse deformation, a multi-angle image acquisition and edge reflection modeling module is constructed, a deep neural network is used to extract weld seam continuity, edge integrity and hole position geometric consistency features, and a structural feature vector is generated; through small sample abnormal modeling and Gaussian mixed model classification, defect type recognition and credibility scoring are completed, and sub-pixel level coordinate labeling of the defect position is realized by combining a Gaussian fitting algorithm, a defect positioning map is output, and traceability analysis and severity grading are carried out based on historical data comparison, so that the application is suitable for industrial online detection and quality closed-loop control.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Intelligent detection method for machining process of automobile decorative plate

    CN119919400A

  • Power transmission line small sample abnormal data generation method and system based on diffusion model and feature migration

    CN120673192A