The detection of weld defect images using shape-from-shading and wavelet denoising methods

The evaluation of defects and damage is very important in industrial radiography and the use of radiographic images in the detection and assessment of defects is very common. The image quality and the interpreter's experience affect the inspection of radiographs and their evaluation. Radiographic images are often very noisy and other methods of processing are required to reveal the defects. Also, as is generally known, the human eye sees objects in three dimensions, which makes the detection of defects on images with depth easier. Shape-from-shading is a useful method for making three-dimensional images from two-dimensional ones. In this paper, a modified shape-from-shading algorithm has been used to extract the weld defect from the radiographic image. Firstly, a denoising algorithm has been applied to radiographic images with different defects and then, in order to detect the defects, the shape-from-shading method (SFS) has been applied to the denoised images. Expert opinion has also been used for evaluation of the results. Experts are of the view that the SFS method is useful in the detection of welding defects and the combination of image processing techniques, and also that the radiography method can produce a very clear image, which can be used effectively in the detailed analysis of weld images.