Defects Detection of Weld Image Based on Mathematical Morphology and Thresholding Segmentation
Authors:
- Tong TongTTTong Tong
- Yan CaiYCYan Cai
- Dawei SunDSDawei Sun
- 1 September 2012
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 21619646,p. 1-4
Abstract
Detection of welding defects is a key part of industrial production. Conventional image processing algorithms can not perfectly capable when weld defects or their background are complicated, for which adaptability of image processing algorithm is critical, efficient algorithm of defects was employed after characteristics of weld is fully analyzed. The first key method is based on mathematical morphology theories to enhance the contrast of small size defects. A new iterative threshold algorithm is used as second kernel for better image segmentation. Experimental results show that the combined method could meet high requirements of visual detection for weld defects.Keywords
This publication has 11 references indexed in Scilit:
- Defect detection in inhomogeneously textured sputtered surfaces using 3D Fourier image reconstructionMachine Vision and Applications, 2007
- Automatic classification of defects on the product surface in grinding and polishingInternational Journal of Machine Tools and Manufacture, 2006
- Application of Visual Servoing to an X-ray Based Welding Inspection Robot2005
- A System for High-Resolution, Nondestructive, Ultrasonic Imaging of Weld GrainsIEEE Transactions on Instrumentation and Measurement, 2004
- The research of defect recognition for radiographic weld image based on fuzzy neural network2004
- Ultrasonic inspection of defects in welded plates and tubes2002
- Morphologic edge detectionIEEE Journal on Robotics and Automation, 1987
- Refined filtering of image noise using local statisticsComputer Graphics and Image Processing, 1981
- Digital Image Enhancement and Noise Filtering by Use of Local StatisticsIEEE Transactions on Pattern Analysis and Machine Intelligence, 1980
- A Threshold Selection Method from Gray-Level HistogramsIEEE Transactions on Systems, Man, and Cybernetics, 1979