Automatic classification of industrial CT image based on SVM

There are always some flaws in the industrial alloy due to various factors in the foundry process. Industrial workers still manually classify these flaws. In this paper, we present a method which can do the classification automatically. Images to be classified under any scan voltage are transformed to a standard voltage based on the multiple linear regression algorithm. Then characteristic vectors are extracted automatically. Finally, the image is classified by discrimination function which is computed based on Support Vector Machine (SVM) method. Experiments show that our method produces a good precision.