Research On Geometric Parameter Calibration And Image Reconstruction Algorithm For CL

Computed Laminography(CL)can provide the three-dimensional image of the object through the projection information,it is widely used in nondestructive testing and materials analysis for its specialization in testing flat objects,such as multilayer printed circuit boards and solar panels.However,the existence of the tilted angle makes calibrating geometric parameters more difficult,and that is easy to cause geometric artfacts.And the loss of projecton data will cause interlayer aliasing.So the research on geometric parameter calibration and image reconstruction algorithm has a very important theoretical and practical value.In view of the shortcomings of the existing methods,this dissertation focuses on geometric parameter calibration,analytic reconstruction algorithm and iterative reconstruction algorithm.The main research is included as follows:1?An iterative geometric complete parameter calibration method is proposed.The existing methods only can calibrate a part of parameters,and that is difficult to meet the needs.In fact,the difference between CL and CT is there is a tilted angle in CL.Firstly,we analyze the efficiency of applying the CT calibration method into CL and determine the sensitive parameter and its preliminary solution.Secondly,a new non-linear least square cost function according to the error between the practical system and the ideal system is presented,thus the sensitive parameter and the other parameters are constantly optimized by the iterative method.Experimental results show that the proposed method can effectively calibrate all the geometric parameters through one calibration,the precision of the sensitive parameter as well as those affected by that parameter are all significantly improved.2?A novel CL filtered back projection(FBP)reconstruction algorithm based on projection transformation is proposed.FBP is suitable for CL when the tilted angle is small,it is quick and the quality of reconstructon image by it is good.But the existing CL FBP algorithms are used through changing the formula of back projection address and ignore the influence of the tilted angle.Firstly,the virtual detector suitable for the FBP algorithm is created,the formula of the projection address in CL geometric structure is deduced.Secondly,the projection is rearranged by projection transform.Lastly,the reconstruction experements show that the proposed method can provide better reconstruction images than the existing CL FBP algorithm.3?A novel algorithm based on the simultaneous algebraic reconstruction technique(SART)and total variation minimization(TVM)is proposed.When the tilted angle is big and the missing data is more,the FBP algorithm will cause interlayer aliasing.Firstly,the TVM objective function under the imaging model is established,SART is used to solve the imaging model and its solution is as the initial value.Secondly,the adaptive steepest descent algorithm is used to solve the objective function.Lastly,the results of reconstruction experiments shows that compared with FBP,the proposed method can suppress the interlayer aliasing causing by the lack of data,and compared with SART,the proposed method can provide better reconstruction images.