Your aim in this topic is to develop an understanding and skills to process data in the spectral domain. The underlying principles, major techniques and dominant issues are considered in some detail and, although it is always possible to take the study to a deeper level, you should be able to make a confident assessment of the information content of multispectral Earth observation data.

By the end of this topic your should be able to

  • compute and interpret histograms of pixel DN, and explain how these distributions are affected by different contrast enhancements,
  • select an optimum three-waveband combination and display that combination as a colour composite,
  • compute and interpret feature space of pixel DN,
  • design strategies for applying Principal Components Analysis to identify the dimensionality of Earth observation data, aid feature selection prior to creating colour composites, and compress data,
  • explain the principles of biophysical modelling and the uses and limitations of empirical (statistical) and physically-based spectral reflectance models,
  • develop a framework for thematic classification and analyse the performance of classifier techniques within that framework,
  • design and implement hard and soft classifications.