Study On Reconstruction Algorithm For Linear Scan Computed Laminography

Computed laminography(CL)is an effective nondestructive testing technology of detecting objects inside information,contrast with the computed tomography(CT),it solves plate-shell objects imaging problems,for example,X-ray don't penetrate and imaging space is limited,CL technique is applicable to nondestructive testing of plateshell objects,such as printed circuit boards,integrated chips,aviation materials.Due to the limitation of CL scanning structure,the projection data is incomplete,and there are finite angle artifacts and aliasing artifacts in the reconstructed images,it cannot meet the requirements of high-precision nondestructive testing of the workpiece.Therefore,it has great practical significance to study CL image reconstruction algorithm and improve the quality of the reconstructed images.This thesis studies the reconstruction algorithm based on the linear CL system,focuses on the CL filtered back projection reconstruction algorithm,analyzes the reconstruction effect of images and proposes improved algorithms.The main research contents are as follows:Firstly,the projection geometric relation of linear CL is deduced,and the analytic reconstruction algorithm of linear CL based on slice direction reconstruction is realized.The experimental results show that the aliasing artifacts of the reconstructed image based on the slice direction are reduced compared with the reconstructed image based on the fault direction.Compared with the linear CL,the projection data of the orthogonal linear CL is more sufficient and the image quality is better.Secondly,a projection view-weighted reconstruction algorithm(VW-FBP)for linear CL is proposed.High density characteristics of non-focal plane will disturb the focal plane reconstructed image,and result in the aliasing artifacts of reconstructed image.By calculating the dissimilarity between different projection angle of slice image,slice images from different projection views are weighted,which can inhibit the interference phenomenon of high density characteristic on the other slice,thus reduce aliasing artifacts effectively.At the same time,the image quality is analyzed by changing the adjustment parameters of the weighted coefficient.The experimental results show that the larger the adjustment parameters of the weighted coefficient,the fewer the aliasing artifacts in the image and the clearer the image,the modulation of slice sensitivity profile is increased by about 0.2.Thirdly,an FBP algorithm based on multidimensional filtering(MF-FBP)for linear CL is proposed.The filter function of FBP algorithm plays an important role in removing the blur of the reconstructed image.In order to eliminate the aliasing between slices,slicedirection filtering is introduced in the filtering process.In this thesis,three kinds of filtering functions are used to analysis the filtering effect.The simulation and actual experimental results show that MF-FBP algorithm can effectively remove the aliasing artifacts in CL image,Hanning window and Hamming window work best,the modulation of slice sensitivity profile is increased by 0.165.Accordingly,the modulation of slice sensitivity profile of Sinc window filtered image is only improved by 0.075 and reconstructed image is over sharpened.