Hyperspectral Image Classification Using Kernel Fukunaga-Koontz Transform

Abstract

This paper presents a novel approach for the hyperspectral imagery (HSI) classification problem, using Kernel Fukunaga-Koontz Transform (K-FKT). The Kernel based Fukunaga-Koontz Transform offers higher performance for classification problems due to its ability to solve nonlinear data distributions. K-FKT is realized in two stages: training and testing. In the training stage, unlike classical FKT, samples are relocated to the higher dimensional kernel space to obtain a transformation from non-linear distributed data to linear form. This provides a more efficient solution to hyperspectral data classification. The second stage, testing, is accomplished by employing the Fukunaga- Koontz Transformation operator to find out the classes of the real world hyperspectral images. In experiment section, the improved performance of HSI classification technique, K-FKT, has been tested comparing other methods such as the classical FKT and three types of support vector machines (SVMs).

Untitled

“…This characteristic yields more effective classification power for the application areas mentioned above. Spectral Signature is nothing but a different type of materials can be represented by a set of bands, this spectral signature simplifies the separation of these materials [14] [15].…”

A supervised discriminant subspaces-based ensemble learning for binary classification

“…KFKT has also proved to be immensely helpful in the field of face detection and recognition [7], helping to resolve difficulties arising out of hyperspectral image classification issues [8,9] as well as the dimensionality reduction of hyperspectral data [10]. Another new technique has been put forward by Binol et al that suggest employing differential evolution algorithm-based kernel parameter selection technique for radial basis function (RBF) kernel within KFKT [11].…”

Kernel Fukunaga–Koontz Transform Subspaces for Classification of Hyperspectral Images With Small Sample Sizes

“…In the literature, Fukunaga-Koontz transform (FKT) is substantially proposed for the two-class problems and generally used for feature selection and ordering [10] and target detection tasks [18] for linearly distributed data sets. Kernelized extensions of the original method were introduced for nonlinearly distributed data sets [15] and multiclass classification problems [17]- [19]. Despite the naming of titles in the referenced papers containing "classification" word, authors used KFKT as a target detection approach and every class is defined as the target and other classes as clutter in turn.…”