Crack detection of industrial CT noisy image
Li Zeng · Optics and Precision Engineering · 2010
A crack detection method of Industrial Computed Tomography(ICT) noisy images based on fast Beamlet transform is presented in this paper. After analyzing the composition and relation of Beamlet in a mono-scale,a fast Beamlet transform is proposed. On the basis of the Beamlet transform,a control variable about the relativity is introduced. Then,combining the tree-structure of Beamlet’s mutiscale and a top to bottom inter-scale inhibition to optimize the object function,the crack is detected. Finally,considering the near pixels of detection result of the crack,the edge of crack domain is extracted. A numerical experiment on the images including an original ICT noisy image,a Gaussian white noise image with a variance of 0.1 and a salt pepper noise image with a superposition density of 0.1 is carried out. Compared with the methods of Laplace,Canny or wavelet,the proposed method can detect the crack of ICT noisy image more effectively. Because the Beamlet transform uses lines to analyze the image data,the proposed method has a robustness to noises.