• Data analysis is very important for signal processing applications
  
  
• PCA is an effective data normalization technique
  
  
• Choice of LDA type is application dependent
  
  
• Class-dependent LDA type proved to be better with an average of 50.41 % error rate over the class-independent LDA type when applied on image database
  
  
• Overall error rate of 39.7% using PCA proved to be better compared to LDA on a 43 feature vector