LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

How to find the shift in pixel between two images

Hello,

 

I am working on Image Correlation and the aim is to find the shift or displacement between two image(pixel shift). I have tried Gwyddion and matlab but I am not getting satisfactory result. I have also tried simple correlation using labview but its bot giving the appropriate result due to noise in the images as they are taken with atomic force microscopy on micro-structure. I would really appreciate for your suggestions.

 

Thank You,

 

Ravi Prakash Singh

 

0 Kudos
Message 1 of 6
(4,885 Views)

If you have the NI Vision Module, there are some nifty object tracking tools that can show you when specific pixel sets move.

 

Otherwise, you're trying to do a pretty complicated algorithm if there's noise involved, so you probably won't be able to get it to work without a sophisticated add-on.

Cheers


--------,       Unofficial Forum Rules and Guidelines                                           ,--------

          '---   >The shortest distance between two nodes is a straight wire>   ---'


0 Kudos
Message 2 of 6
(4,859 Views)

I would think a 2D cross correlation should be quite stable, even in the presence of noise. How big is the shift and how different are the pictures in addition to the shift? Can you show ius a typical pair of pictures?

0 Kudos
Message 3 of 6
(4,847 Views)

Hi Altenbach,

 

Please find the pair to images attached for image correlation and if you have a VI for 2D correlation please share it.

Download All
0 Kudos
Message 4 of 6
(4,841 Views)

....The shift depends with the applied voltage on the structure and the structure is moving  in both x and y-direction along with some noise. These pictures are taken with AFM  (in my case its MEMS comb structure) and the shift is in the range from zero to 50 pixel shift with the varying voltage applied.

0 Kudos
Message 5 of 6
(4,838 Views)

The plain LabVIEW cross corelation accepts 2D arrays directly. Of course you need to either work on a specific color component or convert to grayscale first.

0 Kudos
Message 6 of 6
(4,826 Views)