Image Classification Based on Fuzzy Support Vector Machine

As a basic two-class classifier, support vector machine (SVM) has been proved to perform well in image classification, which is one of the most common tasks of image processing. However, for the n-class problem in image classification, SVM treats it as n two-class problems, in this way, unclassifiable regions exist. In this paper, we introduce fuzzy support vector machine (FSVM) and define a membership function to classify images which are unclassifiable using conventional SVM. For the input vector of SVM and FSVM, we use combined image feature histogram. Being compared with the conventional SVM, FSVM shows the same result as SVM for the images in the classifiable regions, and for those in the unclassifiable regions, FSVM generates better result than SVM.