Features Classification is made on the basis of measured features Features should Be easy (or inexpensive ) to measure or extract Clearly demarcate classes 上 Xamples Medical d agnosis Character recognition □■ DISPLAY CIin) mit 人 16881
Features • Classification is made on the basis of measured features • Features should – Be easy (or inexpensive) to measure or extract – Clearly demarcate classes • Examples – Medical diagnosis – Character recognition DISPLAY( Clin) DISPLAY( Dlin) 16.881 MIT
Feature Vectors Generally, there are several features required to make a classification These features x can be assembled into a vector Any object to be classified is represented by a point in n dimensional feature space mit 人 16881
Feature Vectors • Generally, there are several features required to make a classification • These features xi can be assembled into a vector • Any object to be classified is represented by a point in n dimensional feature space x1 x = x2 x3 x1 x2 x3 x 16.881 MIT
Joint gaussian distribution Density function entirely determined by mean vector and correlation matrix m Curves of constant probabi lty are ellispolds X exp3-(x-m)K(x-m) 2丌 mit 人 16881
Joint Gaussian Distribution • Density function entirely determined by mean vector and correlation matrix • Curves of constant m x2 x1 probability are ellispoids p ( x) = ( 1 ) exp − 1 (x − m )T K −1 (x − m ) 2 K 2 m π 16.881 MIT
Pattern Recognition Model There are two major elements required for pattern recognition a feature extractor A classifier x Raw data Feature Categor extractor Classifier n mit 人 16881
Pattern Recognition Model • There are two major elements required for pattern recognition – A feature extractor – A classifier Raw data Feature Category extractor Classifier x1 x2 xn 16.881 MIT
Template matching Define a template for each class · Choose class based on Maximum correlation or Minimum error What are the limitations? 人 16881 DISPLAY DIin) DISPLAY DIin) mit
Template matching • Define a “template” for each class • Choose class based on – Maximum correlation or – Minimum error • What are the limitations? 16.881 DISPLAY(Dlin) DISPLAY(Dlin) MIT