.png){kind=icon}
返回
Detection Model for Automatic Defect Quantification and Segmentation for Stepped Eddy Current Thermography
.png){kind=icon}
PRE
AI
DOI：10.1109/JSEN.2025.3542091.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
The stepped eddy current thermography (SECT) nondestructive testing (NDT) technique is characterized by long heating time and nonuniform temperature rise. This causes it to remain challenging to quantify the geometrical features of defects on the inner wall of the tank roof and the outer wall of the tank bottom in oil and gas storage tanks. This article proposes a combined model for compressing and reconstructing thermal image sequences: the skewness model combined with the improved Gaussian adaptive background estimation algorithm (SM-IGABEA) for quantifying the defect morphology. The combined model is coupled with the first-order differential max-min method to quantify the width and height of defects accurately. The combined model combined with the first-order differential mean method can accurately segment defects. A mathematical model for predicting the residual depth (RD) of steel plates is developed to describe the relationship between the geometric characteristics of defects and the mean value of skewness. Finally, the generalization of SM-IGABEA is verified by elliptical defects. The results show that various combinatorial models and quantization methods are proposed for the defect measurement task. The measurement accuracy and stability of SM-IGABEA significantly outperform the mainstream compressive reconstruction algorithms.
Keyword:
Steel
Eddy currents
Sensors
Infrared imaging
Heating systems
Temperature sensors
Time series analysis
Surface morphology
Reconstruction algorithms
Oils
Defect quantification study
defect segmentation
nondestructive testing (NDT)
skewness model
stepped eddy current thermography (SECT)
