extraction defects of radiographic images using local information in transform domain

Radiography is one of most common methods for testing the quality of weld. At present radiographic images are interpreted by humans. In order to increase accuracy and speed of interpretation and costs decrease ,We are looking for a way to automatic detect of weld defects. The correct extraction of defects is The most important step to automatic processing.  Therefore ,in this paper a new strategy to extract the weld defects using image processing techniques is proposed.  In the proposed method after segmentation Region Of Interest (ROI), histogram matching and Weiner filter were used to remove noise and enhance the quality of image, then For optimum utilization of image data in different directions, Wavelet transform is used. Afterward by applying techniques such as local thresholding, morphology and edge detection on the components of the wavelet transform defect area of Radiography images was extracted