Probability guided and minimum residual exhaustive searching approaches for subpixel classification

The linear mixing model (LMM) is one of the models to describe the characteristics of mixed pixels in a remotely sensed image. This paper discusses the inverse process, the unmixing process. A minimum residual exhaustive searching (MRES) method and a probability guided (PG) method are presented. The MRES method takes residuals as the measurement to search for the unmixing results with a minimum residual. The PG method makes use of statistical information to determine the dominant components of a pixel and then, to unmix the pixel into the dominant components.