Return to Workshop Listing| Author: | Changjiang Yang |
| Advisor: | Juyang Weng |
| Email: | yangcha1@cse.msu.edu |
Image orientation is important to the automatic image acquisition. We employ the low-level image features to detect the image orientation. Five kinds of classifiers, namely an hierarchical discriminating regression(HDR) tree, a support vector machine(SVM), a RBF network, an LVQ method, and k-NN are considered to determine the image orientations. The comparison among these methods show that the HDR tree and SVM can not only obtain a high accuracy on both training set and test set, but also the classification efficiency is high.