A weld defect detection adaptive preprocessing system and method based on DCM images
By using the DCM image adaptive preprocessing system, dynamically adjusting the combination of mapping parameters and optimizing the MLP model, the problem of reduced contrast in high dynamic range DCM image conversion is solved, enabling effective detection and real-time processing of various weld defects, and possessing self-evolution capabilities.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HANGZHOU DIANZI UNIV
- Filing Date
- 2026-04-15
- Publication Date
- 2026-07-17
AI Technical Summary
Existing weld defect detection systems face challenges in converting high dynamic range (DCM) images into 8-bit visible light images, resulting in reduced image contrast, with tiny defects being submerged in background noise. Furthermore, the different distribution characteristics of different types of defects make it difficult to effectively preserve all defect features.
An adaptive preprocessing system based on DCM images is adopted. Through a multi-parameter mapping module and a multilayer perceptron optimization module, the combination of mapping parameters is dynamically adjusted to generate multiple visible light candidate images with different grayscale distribution characteristics. Combined with image quality analysis and MLP model, the images are optimized or fused to improve detection accuracy and consistency.
It achieves feature preservation for multiple types of weld defects, reduces fluctuations in the missed detection rate, meets the real-time detection needs of industrial production lines, and continuously optimizes the preprocessing effect through a data-driven self-evolution mechanism.
Smart Images
Figure CN122415362A_ABST