Improved Scanning And Approximate Reconstruction Algorithm Of Cone-Beam Industrial CT

Computed tomography (CT) is regarded as the perfect non-destructive test technology (NDT) in medical diagnosis and engeneering. The principle of industrial computed tomography (ICT) is similar to medical computed tomography. Both are based on Radon transform. In ICT, the size, structure and material of the scanned object vary widely. Thus, the mechanical construction, ray energy, imaging precision and image processing in ICT are partially different from that in medical CT. So, the scan way and the reconstruction algorithm should be modified according to the characteristic of ICT. In about fouty years, the scan mode of CT has changed from parallel-beam to cone-beam. Compared with 2D CT, 3D cone-beam CT, in which cone-beam rays are used to scan the object, has much shorter scan time and higher axis resolution because it can make use of the rays more efficiently. So it has attracted increasing attention, and is gradually being used in medical diagnosis and engineering. The image of the object is reconstructed from cone-beam projection data by use of algebraic or analytic algorithms. The algebraic algorithms need iteration, so the implementary time is longer. There are two types of analytic algorithms: exact and approximate algorithm. The loci of exact algorithms must satisfy the Smith's necessary and sufficient condition, and the implemenary efficiency is lower. Although the images quality reconstructed by the approximate algorithms is not as good as that reconstructed by exact algorithm, the efficiency is higher than that of the exact algorithms, and scan locus could be incomplete. Feldkamp, Davis and Kress (FDK) proposed the practical approximate cone-beam reconstruction algorithm for the 3D CT with circular ray-source trajectory, which is a generalization of the fan-beam FBP (Filtered Back Projection) algorithm. Now, FDK algorithm is primary reconstruction algorithm in practical cone-beam CT. When detector size and distance between the source and rotary center were both fixed, we analyzed the condition that any point in the object was scanned and the relation between the cone angle and the complete reconstruction height. In addition, we discussed the space-between of multi-cycle locus. The simulations validate reconstruction images qualify is reduced and the complete reconstruction height become lower when the cone angle is larger.In ICT, it is common that the diameter of the object is larger than the width of the detector. The conventional 3rd generation circular trajectory and FDK algorithm are limited. In this dissertation, a new"2nd+ 3rd"generation scan mode and an off-center reconstruction algorithm were presented for solving this problem. This"2nd+ 3rd"generation scan mode shares similarities with the conventional 2nd generation cone-beam CT, which is traditionary scan way of inspecting the large object. But the"2nd+ 3rd"generation cone-beam CT needs the source and the detector are only translated a few times at each graduation, and the rotary graduation equals that of the 3rd generation cone-beam CT. The object is reconstructed by joining the pieces of the reconstruction area and using an off-center FDK algorithm that is an extension of the standard FDK algorithm. According to the reconstruction method, the projections do not need rebinning and interpolating, so the implementary efficiency is higher. Futhermore, the view of field and translated times are adjustable according to the size of the oject and horizontal flare angle of cone-beam. In simulation experiments, compared with the 2nd generation cone-beam CT, sampling time of this new scan mode is much shorter than that of the 2nd cone-beam CT, relatively redundant ratio is also reduced greatly. The results of simulations show that the images reconstructed by this method are of good quality.In cone-beam CT, because of the cross talk between detector units caused by X-ray's scatter (especially high energy X-ray's scatter in industrial CT), the Signal-to-Noise of detector and the space resolution of reconstructed image are decreased, and the artifact near the edges of reconstructed image is also generated. We discussed a sample method. The heavy metal partitions are fixed on the detector horizontally and vertically between the adjacent detector units, or are fixed in front of the detector. The object is scanned four times, and the projections are accordingly sampled four times. This sample way could decrease the cross talk of X-ray, and projections are complemented. After each scan, cone-beam projections are backprojected using FDK algorithm respectively. From the results of simulation experiments, we see the presented sample method can efficiently reduce the artifact caused by cross talk of X-ray's scatter, and keep the space resolution of reconstructed images.