The output of IMAQ correlate is completely wrong

[MSleiman]

Hey Evan D.

I am using the IMAQ Correlate VI in the LABVIEW Student Edition 2009.

and i have a 'little' two questions for you:

1> The above mentioned VI outputs an image scaled from 0 till 255, depending on the degree of matching; my question is: how did they obtain these values,

i used the formulas attached and i obtained values from -1 to 1.

2>where bugs fixed in IMAQ Correlate VI (bugs mentioned by Root Canal ) in LABVIEW 2010 Student Edition.

Best Regards

M. Sleiman

Electrical Engineer

YL-Engineering

[rpursley8]

1.  The output is just rescaled from [-1, 1] to [0, 255]

2.  The bugs seem to be fixed.  A direct method and the IMAQ Correlate produce very similar but not exact results now.

[MSleiman]

Thanks Randall

[MSleiman]

Here's my VI's....

thanks RC.

M.Sleiman

[tgmiller]

I would debate whether this VI has been fixed yet or not.  I am enclosing a VI where I take an array and a kernal, convolve them using the Convolution vi, then convert the array to an IMAQ and again convolve with the kernal but using the IMAQ Convolute vi, yielding a notably different result.  I would be interested to hear if there is a bug in my code or if it's still in the IMAQ Convolute vi.

Teresa

[tgmiller]

Just noticed this threat was about correlate and not convolute.  I suspect there is a problem with the 2d convolution IMAQ vi though.

[Eric-J]

Hi tgmiller,

What versions of LabVIEW, Vision Acquisition Software, and Vision Development Module are you using?

[tgmiller]

I'm using Labview 2010 SP1, Vision Dev 2010 SP1 and, Vsion Acq August 2010.  

My coworker believes the problem is likely to be something to do with how the borders are treated by each convolution.  If that is the case then I wish there were better documentation on these two functions because I really think that a 2D conv that takes the same inputs and returns an output of the same sizes should give the same result unless clearly stated why it will not in the documentation.  

Thanks,

Teresa

[Eric-J]

Thank you for providing that information. I will file a bug request for this and forward it to our R&D team.