Weld defect features extraction on digital radiographic image using Chan-Vese model

Accurate inspection of welded materials is important in relation to achieve acceptable standards. Radiography, a non-destructive test method, is commonly used to evaluate the internal condition of a material with respect to defect detection. The presence of noise in low resolution of radiographic images significantly complicates analysis; therefore attaining higher quality radiographic images makes defect detection more readily achievable. This paper presents a study pertaining to the quality enhancement of radiographic images with respect to different types of defects. A series of digital radiographic weld flaw images were smoothed using multiple smoothing techniques to remove inherent noise followed by top and bottom hat morphological transformations. Image quality was evaluated quantitatively with respect to SNR, PSNR and MAE. The results indicate that smoothing enhances the quality of radiographic images, thereby promoting defect detection with the respect to original radiographic images.

Download Full-text

Development of Radiographic Image Classification System for Weld Defect Identification using Deep Learning Technique

International Journal of Scientific and Engineering Research ◽

10.14299/ijser.2021.05.01 ◽

2021 ◽

Vol 12 (5) ◽

pp. 390-394

Author(s):

Distun Stephen ◽

Dr.Lalu P.P

Keyword(s):

Neural Network ◽

Deep Learning ◽

Neural Network Model ◽

Classification System ◽

Radiographic Image ◽

Radiographic Images ◽

Defect Identification ◽

Weld Defect ◽

Learning Technique ◽

Welding Defect

Weld defect identification from radiographic images is a crucial task in the industry which requires trained human experts and enough specialists for performing timely inspections. This paper proposes a deep learning based approach to identify different weld defects automatically from radiographic images. To employ this a dataset containing 200 radiographic images labelled for four types of welding defect- gas pore, cluster porosity, crack and tungsten inclusion is developed. Then a Convolutional Neural Network model is designed and trained using this database.

Download Full-text

Automatic weld defect detection based on potential defect tracking in real-time radiographic image sequence

NDT & E International ◽

10.1016/j.ndteint.2011.10.008 ◽

2012 ◽

Vol 46 ◽

pp. 14-21 ◽

Cited By ~ 25

Author(s):

Jiaxin Shao ◽

Dong Du ◽

Baohua Chang ◽

Han Shi

Keyword(s):

Real Time ◽

Defect Detection ◽

Image Sequence ◽

Radiographic Image ◽

Weld Defect

Download Full-text

Digital Radiographic Image Enhancement for Weld Defect Detection using Smoothing and Morphological Transformations

Scientific Research Journal ◽

10.24191/srj.v9i1.5053 ◽

2012 ◽

Vol 9 (1) ◽

pp. 15

Author(s):

Suhaila Abdul Halim ◽

Arsmah Ibrahim ◽

Yupiter Harangan Prasada Manurung

Keyword(s):

Defect Detection ◽

Test Method ◽

Radiographic Image ◽

Radiographic Image Enhancement ◽

Radiographic Images ◽

Weld Defect ◽

Internal Condition ◽

Different Types ◽

Morphological Transformations ◽

Non Destructive

Accurate inspection ofweldedmaterials is important in relation to achieve acceptable standards. Radiography, a non-destructive test method, is commonly used to evaluate the internal condition ofa material with respect to defect detection. Thepresence ofnoise in low resolution ofradiographic images significantly complicates analysis; thereforeattaining higher quality radiographic images makes defect detection more readily achievable. This paper presents a study pertaining to the quality enhancement of radiographic images with respect to different types of defects. A series of digital radiographic weld flaw images were smoothed using multiple smoothing techniques to remove inherent noise followed by top and bottom hat morphological transformations. Image quality was evaluated quantitatively with respect to SNR, PSNR andMAE. The results indicate that smoothing enhances the quality ofradiographic images, thereby promoting defect detection with the respect to original radiographic images.

Download Full-text