Study On Defect Detection Algorithm Of Industrial CT/DR Images Based On Wavelet Transform And C-V Model

Industrial Computed Tomography(ICT) and Digital Radiography(DR) are two important technical about nondestructive detection. The defects which are contained in the workpiece to be detected can be obtained by scanning. Defects detection is the key to segmentation and measurement on CT/DR images. It plays a crucial role to ensure the safety and reliability of workpiece. In order to analyze and measure the defects, it is necessary to exact the defects from the images, which is the basis of the following work.Suffering from the limitation of the imaging conditions, some image data include noises or vague edges or uneven background brightness and so on, traditional image segmentation methods can not obtain accurate results. C-V model which is on the basis of curve evolution theory and level set method can obtain the better results. Because of its time-consuming, the speed of evolution is very slow. In order to tackle these problems, an improved method will be adopted. Wavelet transform will be applied in the C-V model to improve the traditional methods. Firstly, wavelet transform will be applied on CT/DR image segmentation. After that, a set of multiscale image set will be obtained. The contour of the coarse scale will be firstly found by C-V model. Then the result will be interpolated into the fine scale regarding as the original contour of the evolution. The improved method can not only enhance the speed of evolution, but also reduce image noises.In view of the properties of DR images and the slow speed evolution on C-V model, the method based on wavelet transform and C-V model will be presented in this paper. Because the evolution result of coarse scale is close to the actual edge, it is favourable to get the result when it is interpolated to the fine scale. Following this idea, we can obtain the contour of the original DR image. The result of the experiment shows the efficiency and reliable of the method.In order to improve the precision of the evolution result on C-V model, it is necessary to study a sub-pixel measuring method. On the basis of the result of wavelet combining with C-V model, linear interpolation will be used to make the edge located in sub-pixel level. The experiment results show that the method outperforms other methods.3D Industrial Computed Tomography Image which contains cracks can be obtained by scanning. Exacting the cracks from CT image is of enormous significance to analyze its property and ensure the security of the workpiece. This paper researches a method which combines 3D wavelet transform with C-V model to improve the speed of crack detection. The idea of this method is the same as 2D wavelet combining with C-V model. This hybrid method has advantages of both 3D wavelet transform and C-V model. It is speedy to detect the cracks inside the ICT images without of considering the priori information such as the direction, the shape or the position of the crack. The detection results demonstrate that our method can provide a feasible solution for practical application.