3D Digital Tomography Image Reconstruction to Determine the Dimensions of Discontinuities in the Material Using MATLAB

To acquire more accurate information, such as the existence of discontinuities on an object from 2D digital radiography images, a reconstructed 3D image using a tomography technique is needed. In this paper, an algorithm was developed to determine the discontinuity dimension. The object used in this study is made of gypsum, which has three types of artificial discontinuities: a 4.80 mm iron ball, an 8.00 mm iron ball, and drill holes. The processes to get projection data consist of preprocessing to convert the images from red-green-blue to grayscale image format, segmentation to differentiate the object from the background using the Thresholding method and Active Contour Chan-Vese model, and morphological operations to visualize the image into 3D volumetric, followed by determining object dimensions, and subsequently the reconstruction results. The tomographic image reconstruction was constructed from 90 images irradiated by an x-ray machine using digital radiography with constant irradiation parameters and a 2° rotational angle increment interval from 0°-180°. The error results of 4.80 mm iron balls discontinuity are 1.5%, 8.00 mm iron balls are 4.1%, and drill holes are 7.05%. Moreover, the Misclassified Area-Mutual Overlap (MMO) method is employed to test the segmentation results, which resulted in an MMO value of 78.23%.