焊接
横截面(物理)
分割
灰度
计算机科学
特征(语言学)
人工智能
特征提取
数学形态学
材料科学
有孔小珠
图像分割
图像处理
计算机视觉
模式识别(心理学)
像素
图像(数学)
复合材料
物理
量子力学
语言学
哲学
作者
Ting Lei,S. L. Gong,Chaoqun Wu
出处
期刊:Materials
[Multidisciplinary Digital Publishing Institute]
日期:2024-09-24
卷期号:17 (19): 4683-4683
摘要
In the field of welding detection, weld bead cross-section morphology serves as a crucial indicator for analyzing welding quality. However, the extraction of weld bead cross-section morphology often relies on manual extraction based on human expertise, which can be limited in consistency and operational efficiency. To address this issue, this paper proposes a multi-layer multi-pass weld bead cross-section morphology extraction method based on row–column grayscale segmentation. The weld bead cross-section morphology image is pre-processed and then segmented into rows and columns based on the average gray value of the image. In order to extract the feature of multi-layer multi-pass weld feature images, an outline showing the binarization threshold is selected for each segmented image (ESI). Then, the weld contour of ESI is extracted before image fusion and morphological processing. Finally, the weld feature parameters (circumference, area, etc.) are extracted from the obtained weld feature image. The results indicate that the relative errors in circumference and area are within 10%, while the discrepancies in maximum weld seam width and maximum weld seam height can be close to the true value. The quality assessment falls within a reasonable range, the average value of SSIM is above 0.9 and the average value of PSNR is above 60 on average. The results demonstrate that this method is feasible for extracting the general contour features of multi-layer multi-pass weld bead cross-section morphology images, providing a basis for further detailed analysis and improvement in welding quality assessment.
科研通智能强力驱动
Strongly Powered by AbleSci AI