Study On X-ray Image Filtering And Enhancement Methods For Integrated Circuit Packaging Detection

With the development of fabrication technology in integrated circuits, the requirementsfor the internal defect detection of the packaged components are getting higher and higher.The traditional defect detection technology cannot detect the internal defects of the packagedcomponents. According to transmission theory, X-ray can go through the packagedcomponents and form images so that we can observe the internal defects of the packagedcomponents.The detection objects of integrated circuit are much smaller than the traditional detectionobjects. In order to achieve detection, they can only be imaged by the micro-focus X-ray.However, the flux of the X-ray tube is very low in the micro-focus X-ray imaging process,which causes serious noise in images. Meanwhile, because the packaged components arealmost thin chips, the passing distance of the X-ray is very short, the exiting attenuation rate issmall and the images have low contrast. Therefore, considering the micro-focus X-ray imagesin integrated circuit detection, this paper has completed the following research about studyingon the image filtering and enhancement:1. The principle of integrated circuit package detection is firstly presented. Through theanalysis of the generating process of the X-ray images, the reason why the X-ray images havestrong noise and low contract is obtained and the basis why putting forward the mixed noisemodel is provided.2. Considering the characteristic of the strong noise in the integrated circuit X-rayimages and the problem that the traditional filtering method cannot keep the image contrastinvariant while filtering, an X-ray image homomorphic filtering method based on wavelettransform is proposed. It can effectively filter the noise of the integrated circuit X-ray images.3. The integrated circuit X-ray images have low contract and they will become fuzzyafter filtering. As the pre-existing methods aren’t able to prominently display the image detailswhile the global images are enhanced, a method about multi-scale decomposition ofLaplacian Pyramid is put forward to enhancing the details of the X-ray images. According tothe result of the experiments, this method can effectively enhance the X-ray image details sothat packaged components internal defects are detected well. 4. Micro-focus X-ray images have the characteristics of the sparsity and mixed noise,but the traditional compressed sensing methods only consider removing additive noise and itis easy to produce "artifact" when dealing with the low contrast image. To solve this problem,a new restoration method based on compressed sensing and total variation is proposed.Experimental result shows that compared with traditional methods, this algorithm not onlycan remove the mixed noise effectively, but also avoids the problem of "artifact".