Active Contour Model Based Image Segmentation Methods And Its Application On Industrial CT

Computer tomography technology uses X-rays to scan the object and then reconstructs the tomographic image of the object based on the obtained projection data.In industry,CT technology has been widely used in many areas of non-destructive testing and non-destructive evaluation and reverse engineering since it is contactless and non-destructive.Image segmentation is an important step in the post-processing of industrial CT images.Based on the target segmentation and extraction,visualization,measurement and other follow-up operations can be performed.After years of development,image segmentation technology is becoming more and more mature.Active contour model based segmentation is a class of method base on PDE(Partial Differential Equations),These methods combine the gray information of the image with the overall information of object contour when modeling,and eventually formed an energy functional with some constrains,and it segments the image by minimization the energy functional with some tools,such as level set method.It not only has higher positioning accuracy,but also can get closed boundary.Because of its good segmentation performance and relatively perfect theoretical basis,it is widely used in medical and industrial.In this paper,we first introduced the research background and significance,and the traditional image segmentation methods,the algorithm based on active contour model and its research status are emphatically introduced.Next,we introduced some necessary theoretical basis of the paper,including curve evolution theory,level set method and variation principle.Then,aiming at the problem of defect segmentation in industrial CT,a level set defect segmentation method based on L0 smoothing and main part localization is studied.The image is smoothed while maintaining the image boundary information at first,then the 2D RCV(Robust CV Model)method is used to locate the main part of the CT image,and finally based on the PLEACM(p-Laplace Equation based Active Contour Model)method,we limit the calculation on the main part,the main part localization image segmentation was achieved.Compared with the original PLEACM method,the experimental results and error evaluation indicate that the method is better and can be applied to actual industrial CT images.In addition,the 2D RSKM model has good robustness and can handle many types of images,we extended the 2D RSKM model to 3D and applied it on several real CT images with good results,by comparing with the 3D RCV model,experiments show that 3D RSKM performed better and adapted more types of data.