A Novel SVM for Crack Edge Checking of Coal CT Image

After analyzing the characteristic of ICT (industrial computerized tomography) image and wavelet transformation singularity, an crack edge checking method for coal CT image based on SVM was studied and improved, and applied in crack edge checking of Coal CT image. First we get the direction and approximate bound of the crack using SVM, and then compare the average area grayscale to locate the crack area. At last, we adopt LS-SVM to detect the crack edge, which is processed using polynomial fit to get an accurate, continuous and independent crack edge image. The experiment of Coal CT image rotation validates the robustness of this algorithm, and the comparative experiments with other method prove that the algorithm has strong robustness against noise.