Research On Segmentation Algorithm Of Industrial CT Image Using Improved Active Contour Model

Industrial computed tomography(ICT) has been widely used in petroleum, materials, railway, aerospace and other fields. By getting the scan information of the target object, ICT could reconstruct the tomographic image of the target object, and then use tomographic image for further analysis and processing. Wherein the key of ICT application is how we can stably, efficiently and accurately abstract the target region form ICT image that we interested. However, there already exist many image segmentation methods, wherein the active contour segmentation model has been widely attracted people's attention because its smooth and closed contour segmentation.In this paper, we deeply study the image segmentation method which based on the active contour model. Based on the exiting active contour models, for the reason of some characteristics of ICT for image defects, in order to improve the accuracy of segmentation, reduce the computing time and increase the initial contour selection robustness, we put forward an improved model of image segmentation algorithm RSF. The main contents of this paper are:(1) Analyze the RSF model parameters and initial contour impact on the segmentation results by experiment, and we draw a conclusion on top of this impact: in the early stages of initial contour segmentation we could increase parameters to reduce the value of the initial contour model for sensitivity to iterative away from the target area, and in the end of contour segmentation we could reduced the iteration parameters so that the active contour closer to the object boundary and reduce the computation time of conclusion.(2) This paper proposed a larger RSF parameter model to replace the original weighting and local and global. on the basis of LGIF model.(3) On the basis of the variation of parameters previously we proposed an adaptive algorithm so that the parameters can dynamically update its value during the evolution of the curve.In order to verify the practicality and effectiveness of the algorithm proposed in this paper, in the fourth chapter, we use some classic uneven gray image segmentation for contrast, and in the fifth chapter we study some defective image segmentation of ICT. In order to analyze the accuracy of quantitative division, in the fifth chapter we introduces two evaluation parameters segmentation to analyze the accuracy and quality, and then quantitatively evaluated segmentation results. The results of experimental show that for defect segmentation of ICT image, the model that we proposed in this paper is better than the RSF model and LGIF model, the proposed model can improve the segmentation accuracy, reduce the computation time and the sensitivity of the model to the initial outline. Therefore, we could get the conclusion that the proposed model in this paper could more better satisfy accuracy and time to the defect segmentation of ICT image.