A New Shaped Fiber Classification Algorithm Based on SVM

Fiber classification, especially shaped fiber classifi-cation, is always an important area in textile analysis. Traditional manual or semi-manual ways to classify different type of fibers will take a lot of time. Support Vector Machine (SVM) is an efficient and robust classifier that will fulfill the requirement on fiber classification. In this paper, a shaped fiber classification method based on Support Vector Machine (SVM) and Kernel Principal Component Analysis (KPCA) is proposed. The shaped fiber´s features extracted by KPCA are used to train and test SVM for obtain suitable parameters of SVM. The experimental results show that our presented algorithm is efficient and robust on classifying shaped fibers.