0

I am trying to make a program that classifies image into various component like detecting the land, water & sky area. I am at a phase where I have separated RGB values of different area into different classes and stored it in a text file (See Image) and calculated its mean and variance.

But I am not able to understand how to implement Maximum Likelihood Classifier so I can move forward to classify image.

Help would be appreciated. Thanks.

Amro
  • 123,847
  • 25
  • 243
  • 454
  • Logistic Regression is essentially a Maximum Likelihood Classifier. See the duplicate for more details. Don't let the title fool you. The OP has working code, but there are a few things in the accepted answer that are required to get it to work properly. – rayryeng Mar 27 '16 at 07:42
  • @rayryeng hmm, while Logistic Regression and Maximum Likelihood Classifiers (think Bayesian Decision Theory) are related, they are formulated and trained differently (one is a discriminative classifier, the other generative). Think `fitglm` vs. `fitcdiscr`.. – Amro Mar 27 '16 at 08:13
  • @Amro you're right. In fact Bayesian Decision Theory was the thing I was leaning towards but I couldn't find a good duplicate. Do you have one in mind? I also thought the third column was the class number so I figured it may have been a good duplicate. – rayryeng Mar 27 '16 at 08:19
  • not really no. Personally I always go back to the book "Pattern Classification" by Duda and Hart whenever I need to refresh on Bayesian Decision Theory... Perhaps I can point OP to to online resources like this one: https://www.byclb.com/TR/Tutorials/neural_networks/ch4_1.htm – Amro Mar 27 '16 at 08:23
  • @Amro it's a classic book. I use that for Bayesian Decision Theory all the time. Good resource! – rayryeng Mar 27 '16 at 08:28

0 Answers0