imcomplement for binary image doesn't replace 1.000 with 0.000

I have a binary image that I am certain has only values of 0.0000 and 1.0000. I need the complement of this image, but when I use imcomplement, all the values that are 1.0000 become very small nonzero numbers (eg 1.0245e-16) and this ruins the rest of my program b/c I need to use bwdist later. Am I doing something wrong?

[Jan]

Usually a binary image can be stored as LOGICAL or UINT8 array to save memory. This would solve the floating point problems implicitely. Therefore it is called *binary*.

Answer 1

[Friedrich]

Hi,

are you sure that your value is really 1.0? Because:

   imcomplement(.9999999999999994)

ans =

    5.5511e-016

>> .9999999999999994

ans =

      1.0000

For varifying you can round your values and take a look at the result:

     imcomplement(round(.9999999999999994))

ans =

       0

[Jan]

+1. It seems to be another floating point problem.

[Andrei Bobrov]

double(int8([1 0 1.12e-14]))