From Subpixel to Superpixel: A Novel Fusion Framework for Hyperspectral Image Classification

Abstract

Supervised classification of hyperspectral images (HSI) is a very challenging task due to the existence of noisy and mixed spectral characteristics. Recently, the widely developed spectral unmixing techniques offer the possibility to extract spectral mixture information at a subpixel level, which can contribute to the categorization of seriously mixed spectral pixels. Besides, it has been demonstrated that the discrimination between different materials will be improved by integrating the geometry and structure information, which can be derived from the variance between neighboring pixels. Furthermore, by incorporating the spatial context, the superpixel-based spectral–spatial similarity information can be used to smooth classification results in homogeneous regions. Therefore, a novel fusion framework for HSI classification that combines subpixel, pixel, and superpixel-based complementary information is proposed in this paper. Here, both feature fusion and decision fusion schemes are introduced. For the feature fusion scheme, the first step is to extract subpixel-level, pixel-level, and superpixel-level features from HSI, respectively. Then, the multiple feature-induced kernels are fused to form one composite kernel, which is incorporated with a support vector machine (SVM) classifier for label assignment. For the decision fusion scheme, class probabilities based on three different features are estimated by the probabilistic SVM classifier first. Then, the class probabilities are adaptively fused to form a probabilistic decision rule for classification.

Existing System

Multinomial Logistic Regression (Mlr), Hyperspectral Image (Hsi) Classification, Superpixel-Based Discriminative Sparse Model.

Proposed System

In this paper, novel frameworks to adaptively integrate the subpixel-, pixel-, and superpixel-based complementary information are proposed for the HSI classification. Here, both the feature fusion and decision fusion schemes are considered. First, the subpixel-level structure feature, pixel-level spectral mixture feature, and superpixel-level spectral–spatial feature are extracted, to characterize varying materials from different aspects. In the feature fusion scheme, three different kernels that are induced from the considered features are added up to generate a composite kernel, which is then incorporated with an SVM classifier to determine the labels. By this way, each specific feature as well as the cross information between different features can be exploited to improve the discrimination capability. In the decision fusion scheme, each feature is used to estimate the initial class probabilities for each class. Then, the fusion of class probabilities is guided by the CDCPs per each pixel and confidence scores of different classifiers, leading to the final estimation of classification results. With the introduced decision fusion scheme, three types of available features are adaptively integrated for higher classification accuracies.

Architecture

Feature Fusion based Classification

Decision Fusion based Classification