PDE-Based Model for Weld Defect Detection on Digital Radiographic Image

Partial differential equation (PDE)–based image processing has played a substantial role and become more popular in the recent years. In the application of weld defect detection, the PDE models can be applied for image smoothing and segmentation. In this study, anisotropic diffusion proposed by Perona Malik known as Perona Malik Anisotropic Diffusion (PMAD) model is used as a denoising process for smoothing while level set by Chan and Vese is used as detection process for segmentation. The PMAD model has been solved using Peaceman Rachford (PR) scheme in order to improve the denoising process. A set of radiographic images that contain weld defects are used as input data. The implementation of the algorithm is done using Matlab R2009a. The average error on contour-based metric and CPU times are used to evaluate the accuracy and the efficiency of the CV model and the thresholding method on the proposed denoising process. From the results, the contour detection of weld defect is improved on image after denoising process using CV model as compared with the thresholding.In conclusion, the PDE-based model can be applied in detecting weld defect on radiographic images which could assist radiography inspector in their inspection for an accurate evaluation.