A simple decision rule for remote sensing and digital image classification is the ParallelPipe or Parallelepiped classifier, used to assign pixels to specific classes.
This technique establishes decision boundaries as multi-dimensional rectangles (parallelepipeds) within feature space.
Each class is defined by a set of value ranges for each spectral band; the classifier verifies if a pixel's spectral values lie within these defined ranges.
Pixels whose spectral values fall within these ranges are then allocated to the corresponding class.
Despite its ease of implementation, this method may result in unclassified or ambiguously classified pixels when spectral ranges overlap.
Consequently, ParallelPipe represents a foundational yet elementary approach to image classification.