Spectral and Spatial Indices based Specific Class Identification from Airborne Hyperspectral Data

Project: Non-funded project

Project Details


Hyperspectral remote sensing has emerged as one of the most versatile and information rich source of data. Hyperspectral data finds applications in various domains, like specific land use land cover identification. Therefore, hyperspectral imaging has emerged as one of the most significant research areas. Hyperspectral data has many advantages, but the major limiting disadvantage to its applicability is that of its dimensionality. Known as the “Hughes Phenomenon”, traditional classification and image processing approaches fail to process data along many contiguous bands due to inadequate training samples. Another aspect of classification is to deal with the real world scenario of mixed pixels i.e. presence of more than one class within a single pixel. An attempt has been made to deal with the problems of dimensionality and mixed pixel, with an objective to improve the accuracy of class identification.

The present research intends to use an indices based approach to deal with the dimensionality of the Airborne Prism EXperiment (APEX) hyperspectral dataset and improve the classification accuracy using Possibilistic c-Means (PCM) algorithm. The APEX Open Science Dataset (OSD) was acquired over Baden, Switzerland and consists of 285 bands at 1.8 m resolution, allowing development of spectral and spatial indices to describe the information in the dataset in a lesser dimensionality. This reduced dimensionality is used for classification, attempting to improve the accuracy of determination of specific class of interest, in the context of a number of classes. Spectral indices are derived from the features of the dataset and spatial indices have been defined using texture analysis over defined neighbourhoods. The classifications of 20 classes of varying spatial distributions were considered in order to evaluate the applicability of spectral and spatial indices in extraction of specific class information.

The proposed input related outputs of classification were compared to a traditional dimensionality reduction approach i.e. PCA in this research. In order to understand the effect of indices of classification, spectral and a combination of spectral and spatial indices were individually assessed and studied. Improvements in specific class determination were assessed by certainty measures of entropy and correctness measures of user’s and producer’s accuracies. Besides reduction of entropy for classes across the various approaches, an overall classification accuracy of 80.50% was obtained for spectral-spatial indices input database against 65% (spectral indices) and 59.50% (optimally determined principal components).

Utilizing the high dimensionality and fine resolution of the airborne hyperspectral dataset, spectral response knowledge – based approach to feature selection, extraction and reduction of dimensionality has been researched. An improvement in accuracy of class determination supports the motivation of the research, concluding that the utilization of the indices and their spectral – spatial properties can indeed improve in identification of land use/land cover on the Earth.
Effective start/end date2/09/1320/03/14

UN Sustainable Development Goals

In 2015, UN member states agreed to 17 global Sustainable Development Goals (SDGs) to end poverty, protect the planet and ensure prosperity for all. This project contributes towards the following SDG(s):

  • SDG 2 - Zero Hunger
  • SDG 11 - Sustainable Cities and Communities


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.