The Main.m is a program which compute traditional evaluation metrics and statistically-based metrics for classifiers selection. For correct work - indicate all of folders and subfolders in path!

The program performs next stages (in the loop - for every input image):

1 - open the test image from graphics file;

2 - apply 10 binarization algorithms for every image;

3 - Calculate and print the table with applied traditional metrics for binarization algorithms, according to Ground Truth;

4 - Calculate and print the table with applied statistically-based metrics for binarization algorithms (without Ground Truth);

5 - Calculate the order of classifiers from best to worsed by using Reference Method;

And, in the end:

6 - Calculate how good not-traditionally used evaluation systems (Pseudo-metrics and proposed Reference method ) are good by calculating (average values through dataset):

• Average sequence alignment cost between the order of classifiers
• Average word-edit distance between the order of classifiers
• Average correlation of algorithm`s order between the order given by F-Measure and statistically-based approaches