Methods of Improving the Quality of X-Ray Image Reconstruction with Limited Projection Data

摘要

Nowadays X-ray Nondestructive Testing (NDT) has played an important role in defect inspection but the X-ray image contains all the information the rays passed by, therefore, it is hard to inspect the inner defect of PCB correctly. The correct inspect is rely on an image reconstruction system. In this study, a X-ray image reconstruction algorism to inspect the defect in a Printed Circuit Board (PCB) is developed. The Algebraic Reconstruction Technique (ART) is used to reconstruct the vertical section image from the images of several different project angle, the images are from the Planar Computer Tomography (PCT) system for PCB. The X-ray is assumed to be parallel beam. The limited angle of projection in PCT reduces the image reconstruction quality. To improve the quality of reconstructed image, a method combines the binary steering mechanism with ART to get the correct convergence and avoiding the local optimum solution of ART is proposed. Furthermore, the computational algorism to increase the reconstructed speed and discussing the influence between the image and the noise of projection data are also discussed.