Research On Automatic Detection Of Element In The CT Image Of PCD

Industrial CT (Industrial Computed Tomography, ICT) which can obtain non-contact andnon-destructive information of the PCB internal structure is considered one of the bestnon-destructive detection methods. Printed circuit board, referred to as PCB, is widely used inindustrial products in a variety of electrical applications. With the development of the PCBtechnology, the traces and vias on the PCB as the elements to reflect the relations of the circuitdistribute increasingly, densely and complicatedly. How to detect the elements on the PCB baseon the CT image automatically, accurately and reliably, becomes a desiderated issue aboutimproving the efficiency of non-destructive detection. Therefore, research on automaticdetection of Element in the CT Image of PCB has important theoretical significance andapplication value.In this paper, the automatic detecting the traces and vias in the CT images is studied inallusion to the problem of duplicate and missed detection in the automatic detection of Elementin the CT Image of PCB, This work which goes deep into analysis about CT image data of thePCB characteristics, is based on the gray level difference between the elements and the substrate,and morphology characteristics of the elements. The main contents and innovations are asfollows:1. For the problem that the number of vias is huge, the target is so small and themorphologic is abnormity，an algorithm based on regional roundness is designed to detect thevias in the CT image of PCB,which combine the morphologic character of via in the3D space.First, superimposed every slice data in the vertical direction of the PCB to enhanced vias target,while reduce non-vias goal.After then a new roundness is defined which is used to recognize theregion of vias, meanwhile get the property of the region from which can calculate the property ofvia. Experiments based on actual data show that the method which is robust has a very highaccuracy rate and a high speed. The veracity shows that the method can be used in the actual viadetection of PCB.2. For the problem that detection accuracy is not high and connectivity is not guaranteed inclassical Hough transforms based on edge information, this paper proposes a new algorithm forwire detection of PCB addressing.The proposed method is applicable to the two-dimensionallayered images of printed circuit board which features conductor’s wide distribution, closelydeployed and specified morphological structure. And the introduction of morphological thinningmethod to the proposed one helps us to refine the target area, meanwhile, obtain the descriptivecurve with one pixel width. Then we calculate the positions of all starting and ending points of all segments making up the wire by the inflection points obtained from the wire’s linked list ofcurvature. The clearly better performance comparing with classical Hough transforms and theaccuracy of wire’s positioning and detecting from a CT images of the two-dimensional layeredPCB is demonstrated by experiments.3. Through the analysis of the morphological characteristics of PCB wire inthree-dimensional space, this paper proposes a method based on skeleton algorithm which isbased on improved distance transform to detecting wire in three-dimensional space,. Theproposed method obtains not only the local extreme points of wire in three-dimensional space,but also the parameters of wire width through the modified distance transform at the first step.Then, we determine the inflection points in three-dimensional space based on the definition ofconnecting distance in three-dimensional space so that we can get the positions of the wire’sendpoints and turning points in three-dimensional space. The results of the experimentsconducted by us prove the availability of directly detecting the wire in three-dimensional spaceby this method. In addition, the proposed approach addresses the problem of wire’s accurate3Dreconstruction, owing to the restorability of distance transforms. And this approach can be usedto detect wire automatically from the CT images of PCB, which reduces workload of people indelamination and rectification.Finally, this work is summarized, and the follow-up work is prospected.