Accuracy and precision computation

I have Ground truth image and the result image(Both of them are binary images, the background is black and the crosses are white). I want to find the precision and values (TP,TN,FP, and FN). How can i calculate these values? if the crosses in the result images are't exactly in the same location in the original one but the difference is very small, can i create "error distance"? i mean if the obtained cross in this range, i can consider it as TP ! Plzzzzzz help meeeeeeeeeee Thank you in advance,

Answer 1

[Mohammad Abouali]

Use confusionmat() to build a confusion matrix or contingency table

[Abdulrahman Mkaies]

Sir, Can i put a distance error for the crosses in the result image ?? I mean, if the result cross around ( 10 pixels ) "for example" i want to put it as tp as well

[Mohammad Abouali]

I cant really help on this, because, I don't what is your image, what do you mean by crosses and ...

[Abdulrahman Mkaies]

Its binary images (the background is black with white dots). The original image (Ground truth) has 10 crosses for example. and I have (result image) the same. I want to compare between them to get TP and TN etc, And if there any "little difference" in the location of the dots in the "results image", i want to consider it True Positive as well... Plz help :)

[Salma Hassan]

[C,order] = confusionmat(...)

return Tp , Fp Fn , Tn what i should use to plot the Roc curve from these number