Finding TP, TN, FP, FN using histc. First Input must be a real non-sparse numeric array (error)

I am finding TP, TN, FP, FN (confusion matrix) of predicted and actual image. smapImg is the predicted/output image and gtImg is the actual/ground truth image.

TP = histc(smapImg(gtImg), 0:255);
FP = histc(smapImg(~gtImg), 0:255);
FN = histc(~smapImg(gtImg), 0:255);
TN = histc(~smapImg(~gtImg), 0:255);

I am getting this error "First Input must be a real non-sparse numeric array" while computing FN and TN, while TP and FP are computing correctely. Please help.

[Steven Lord]

What does the following command display? Show all the columns as any of them may contain information important to understanding why you receive that error.

whos smapImg gtImg

[Nataliya]

Answer 1

[Adam Danz]

The negation (~) is turnning the array into a logical array. As the error message indicates, the first input is expected to be numeric (not logical).

~smapImg(gtImg)

Example

t = uint8(rand(1,10))
class(t)
ans =
    'uint8'
    
t2 = ~t; 
class(t2)
ans =
    'logical'

A reproduction of the error using the logical vector

histc(t2,1:3)
Error using histc
First Input must be a real non-sparse numeric array.