Study on Defect Detection Algorithm of Industrial CT/DR Images Based on Wavelet Transform and C-V Model
|Keywords||Industrial CT / DR Defect detection Image Segmentation Wavelet Transform C-V model|
Industrial computed tomography (Industrial Computed Tomography, ICT) and digital X-ray imaging (Digital Radiography, DR) systems are two important non-destructive testing technology. Workpiece be detected by X-ray scanning, you can get the CT / DR images contain defects within the workpiece. Defect detection of defects in industrial CT / DR image segmentation and measurement reliability plays an important role in ensuring casting. Order to identify defects in the workpiece image, needs to be separated from the image, and on this basis be possible to measure and analyze the defect. Due to the impact of the workpiece material and ray radiation collected interfering signals, and other factors, some image data there is more noise defect edge blur, uneven background brightness, contrast high, the traditional image segmentation methods can not be accurate segmentation of image defects region. CV model based on the theory of curve evolution and level set method can better split image, but because of the segmentation process, constantly re-initialize the level set function and iterative solution of partial differential equations to calculate than larger, leading to the split speed very slow. For this shortcoming, an improved algorithm, wavelet transform with CV model image segmentation algorithm integrated. First CT / DR image wavelet transform using the CV model on a coarse scale image segmentation, and then the segmented results are interpolated to the fine-scale image, as the initial outline of its evolution. Improved algorithm not only improves the speed of image segmentation and noise reduction. This completed as follows: 1. 2D DR images containing defects as the research object, 2D DR images using wavelet decomposition for CV model evolution slow CV model and its coarse-scale low-frequency image defects position, the results of roughly near the target area, then its results interpolated to the fine-scale image the CV model interpolated contour as initial contour evolution of the fine-scale image, turn down, until the original image defect region. DR image of the actual casting experiment results show that the method is feasible and efficient. 2 in order to meet the requirements of the practical problems in the accuracy of measurement results, study the measurement method of a sub-pixel. , Using a linear interpolation method based on wavelet CV model combined with image defects to the edge of the pixel-level positioning to sub-pixel accuracy measurement of sub-pixel defects. Simulation results show that this paper improved accuracy of the method is superior to the traditional method. 3. Cracked 3D industrial CT images as the research object, a three-dimensional wavelet transform image segmentation algorithm combined with CV model, and apply it to the actual three-dimensional industrial CT crack extract to improve the speed of three-dimensional crack split. Three-dimensional wavelet transform segmentation algorithm combined with CV model is a combination of methods to promote the two-dimensional wavelet transform and the CV model, with CV model global optimization Wavelet fast decomposition characteristics, do not need to consider the direction of the crack in the evolution of the shape, location, etc. priori information, you can quickly extract the edge of the workpiece internal crack surfaces in three-dimensional industrial CT images, to lay the foundation for the subsequent crack analysis, has an important significance in practical applications.