Extended Isomap using the kernel Fisher linear discriminant function and the Fisher linear discriminant function

摘要

Way to represent the image for pattern classification, to extend the traditional Isomap using kernel Fisher linear discriminant function Fisher linear discriminant function or (FLD) and (KFLD) for classification. Predicts the geodesic distances of the data points corresponding to the image for pattern classification, extended Isomap method is used as a feature vector geodesic distance pair. This method applies the FLD feature vectors to find the optimal projection direction that maximizes the distance between the centers of the clusters of feature vectors. This method may also be applied to the feature vectors KFLD instead of FLD.