Three Dimensional Image Reconstruction Of Industrial Computed Tomography Based On Continuous Tomograms

The three-dimensional (3-D) image reconstruction of industrial computed tomography (ICT) is a comprehensive technology developed in the base of computer graphics, computer vision and computer image manipulation. It is an important branch of computer visualization technology and plays an important role in non-destructive test (NDT).The realization method of 3-D reconstruction of ICT images is described. Detailed introductions are done on the 3-D rendering methods and interpolation methods, which are the hotspots in 3-D reconstruction of ICT images based on continuous tomograms. The iso-surface patch has shorter calculating speed and better image quality, so it attracts the attention at the current time. Contour-line method, Cuberille plotting method, Marching Cubes(MC) algorithm and Dividing Cubes algorithm are included in the iso-surface patch. The interpolation method consists of gray-scale interpolation and shape-based interpolation. If the intervals between the images are short enough, good reconstructed image can be got with the gray-scale interpolation method. In this paper, the iso-value surface is gotten using MC algorithm at the given threshold in the 3-D volume data field. Combined with the linear gray-scale interpolation method, the 3-D image reconstruction of the object can be realized.The whole work of 3-D reconstruction of ICT images based on continuous tomograms is carried out by computer programming. First, a succession of two-dimensional serial images scanned by CT are displayed in their spatial orders by VB programming. Second, the indirect -volume 3-D reconstruction of CT images is developed by MATLAB programming. After such pre-disposals as gray-scale adjusting, smoothness and sharpness filter, a serial of two dimensional CT images are formed into 3-D volume data. The MC iso-surface patch method is applied to reconstruct the data. In the end, the reconstruction 3-D image can be watched from different viewpoints, and can be divided at any position, and can be transparently displayed.