What is mathematical formula is fully suited to give a decision about matching score?

Question

I need a mathematical formula to give a decision based on given situation :

in my issue, there is a decision must be taken based on the following values where :

• Total extracted feature Number in origin image is X , in my test example X is = (1496)
• Total extracted feature Number in object is Y , in my example Y is (273).

Now starting Matching Process ... then thereis a map function takes values from X and try to find matched features in object ... this function take a values called Z where Z is less than X ( must be) in my example Z = 381 ..

Now Z include two parts where

- N = noise features number where N. 
- M = real feature number where M in my example M= 134. 

i would like to use a mathematical formula that give me a decision based on M value for example :

if score(M) for example > 0.25 then object is there otherwise not.

mu issue in every time when i changed image or object numbers are changes .. if i changed image X is changed , when i changed object Y is changed ..

I’m not familiar with mathematics formula but I would ask if log or others can applied

Answer 1

For me what matters are only the matching features and for all these matching feature evaluating the degree of correlation between the scores.

That is something like $\frac1V\sum\limits_{i=1}^{V}correlation(W_i,Z_i)$

This is a candid approach, but this page Wikipedia : Interclass correlation seems to bring better formulas than mine.