V. Elango,
When you say error, do you mean some margin of error that you would expect to be 0 because the images are identical? Or do you mean that LabVIEW is throwing an error?
Additionally, I would recommend that you avoid sequence structures--especially stacked--for your applications. If you use a enumerated state machine your code will be more readable, flexible, and debuggable. As it stands I would be very hard-pressed to track the flow of your program. Local variables are generally frowned upon as well unless there is a reason that you NEED them, and from what I can tell a shift register would accomplish everything you need in your application.
I did look at where that number is calculated, and it would seem that:
1. You check to see if the pixel values are equal in both images.
2. You convert that to a I16 array, then an I32.
3. You sum the values in that array to get the total number of unmatching pixels, and you multiply that by 100.
4. You use Array Size to get the dimensions of the array.
5. You multiply them to get the total number of pixels.
6. You divide the number of unmatching pixels*100 by the total # of pixels so you get the ratio of unmatched to matched pixels.
7. You check if that's less than 10%.
I think the problem is that you have no tolerance in your comparison. Instead of asking if A[n] == B[n], you should be asking is A[n] == B[n] +/- x, where x is some acceptable tolerance.
Basically, 0x00 and 0x01 are both white, but your program would say that they aren't.
Tanner B.
Systems R&D Engineer - Aerospace, Defense, and Government
National Instruments
