Haralick and Shapiro [Haralick and Shapiro, 1993a], like Sonka, Hlavac and Boyle [Sonka et al., 1993] describe how to use the Bayes' rule in image processing. The use of Bayes' rule requires the knowledge of three parameters:

• A priori probabilities of each class;
• Costs of bad classification T_l'/l ;
  
  
• Densities of probability f_l(x) of each class.

• Estimation of the a priori Probabilities .
• Estimation of the Costs of Bad Classification T_l'/l.
• Computation of the Densities of Probability f_l.
  
• Bayes' Rule Implementation
  
  
• Fusion of Four Spectral Bands MSS4 to MSS7 using Bayes' Rule
  
  
• Fusion of the Spectral Band and the Out-Image Data using Bayes' Rule