Automatic segmentation of industrial radiographic images

Gamma radiography plays an important role in checking the quality of welded components in many industries. Determination of the region of interest (ROI) which contains only the welding area from the whole radiography image is important process for elimination of irrelevant information, leaving only the area of interest in the image. This paper presents proposed method for automatic segmentation (ROI determination) of radiographic image based on image processing techniques. ROI is determined by thresholding the image with a threshold obtained automatically. The size matrix which represents the dimensions of the weld area is automatically calculated. Grayscale, intensity, and their variances of the radiographic images are used and compared from the points of the probability of correct detection and the positive false alarm rate. Morphological operations and wavelet transform have been used for increasing the probability of detection and decreasing the false alarm. The results of the proposed method are comparing with the results of the other common methods. The results show that the proposed method is promising for automatic segmentation of welding from the radiographic image.