1 (edited by bmibonn 2007-06-12 13:24:04)

Topic: Is the Image Analyzer suitable for 3D-interferometry ?

Hello Mr. Vinther,

I just tried to use Image Analyzer for 3D-surface construction of an interferometry stringe map file.

I used only the smooth function to reduce the black fringes... and the result was rather good ... but not perfect.
Which kind of filter has to be used for fringe reduction ?  (Fresnel method ?, FFT ? or Moir

2

Re: Is the Image Analyzer suitable for 3D-interferometry ?

I think the best you can do is using a smoothing filter. I don't really understand how the image was created but perhaps selecting Horizontal filter only in the Smooth dialog will give the best result. You can also rotate the picture to be able to control the direction of the filter relative to the fringes.
If you select the circular filter model (which is actually a line when you select Horizontal) the result of the filter will be a running average over lines in the image.
The Gauss smoothing filter might be the one most comonly used, but again I am not familiar with this was of creating images.

Michael Vinther

> software developer <

3 (edited by bmibonn 2007-06-14 8:39:52)

Re: Is the Image Analyzer suitable for 3D-interferometry ?

Hello Mr. Vinther,

the image was created using a Michelson microscope interferometer, which creates only  fringe images.  Afterwards an ASCII-file was created with a fringe analyzing program for interferometry. So, Image Analyzer used the created ASCII file. The ASCII file containg only the points in a grid is showing again the fringes. The ASCII file was smoothed with the SMOOTH Filter of Image Analyzer under Filters function as the other filters are not available when an ASCII file is loaded. After smoothing the 3d image was created using the plugin for 3d generation.

Is the SMOOTH function a bidirectionally working filter ?

The following image is showing the grid lines of the ASCII file when it is created with the interferometry analyzing program:
http://rem-edx.de/grid.jpg      The ASCII file is containing this point grid containing also the   fringes.
The ASCII file can be loaded under http://rem-edx.de/fringe53b.txt

I think the problem is the same when triangulation fringe images are used for the 3d surface creation.
I do know that  phase shifting in interferometry is solving the problem but this would only shorten the gaps between the fringes and therefore it is improving the resolution of images.
It is possible to shift manually the phase and to take a serie of shifted images which are combined using Image Analyzer. The result was showing broadened fringes and thinned gaps between the fringes. I suggest that this effect is the same if you interpolate the space between the fringes.

Is  it possible to overlay a grid over the fringe image using a filter function ?
Is it possible to combine map files of shifted images ?
I suggest a Moir

4

Re: Is the Image Analyzer suitable for 3D-interferometry ?

Hello

Just a short comment to the first part now as I don't have much time, but I will comment on the rest later.
The smooth filter applyed to an ASCII file is bidirectional and the filter shape is a bit strange... If you select Map | Make image then the file will be converted to a normal grayscale image where you have many more filtering options - including better control of the smooth filter.

Michael Vinther

> software developer <

5 (edited by bmibonn 2007-06-15 9:11:57)

Re: Is the Image Analyzer suitable for 3D-interferometry ?

Hello Mr. Vinther,

thank you for the reply.
It seems that you have solved the problems with your advice to convert the ASCII file.
With a user defined filter the 3d visualization was improved.
http://rem-edx.de/blech311x231new3.jpg
All interferometer fringe images are now fringe free and  showing more details on the surface.

Best regards
Weischer

Re: Is the Image Analyzer suitable for 3D-interferometry ?

Dear Mr. Vinther,

I think that it is possible to run a bandpass Filter FFT using Image analyzer (FFT, select tool, clear outside selection, IFFT), but I found a big artefact on top of the image of the inverse FFT. It seems that the FFT-window is longer than the original image .
http://rem-edx.de/artefact.jpg
The area of the original  image is showing the surface with details with an excellent quality and no interferometry fringe is seen.
What is wrong in processing the bandpass FFT ??

Best regards
Weischer

7

Re: Is the Image Analyzer suitable for 3D-interferometry ?

I can see two possible explanations:
1. A bandpass filter with extremely sharp frequency cutoff tends to produce ripple effect. This is because multiplication with a rect function in the frequency domain corresponds to convolution with a sinc function in the spatial domain (See reference). Actually this is why you usually do not use such sharp cutoff filters but instead a more soft transition.
2. Direct FFT only works on images with a width and height that is a power of 2. Therefore Image Analyzer enlarges the image to the nearest power of 2 in both dimensions by adding zeros at the borders. This sinc ripple effect will be most visible in this area. There is a mathematical trick to make it work on any size, but I have no implemented it and I don't know exactly how it works...

Michael Vinther

> software developer <

8 (edited by bmibonn 2007-07-10 16:36:11)

Re: Is the Image Analyzer suitable for 3D-interferometry ?

Sorry, but I cannot follow you. Using the bandpassfilter FFT of the freeware profft.exe (of six norwegian students) there was no ripple effect and no artefact such as depending on a side ratio of 1:2. All details were seen without the fringes of the interferometry.
Please see the following image
www.rem-edx.de/Bandpass-fft.jpg

Best reagards
Weischer

9

Re: Is the Image Analyzer suitable for 3D-interferometry ?

I don't know profft but I guess it does not use a perfectly sharp frequency cut off. If it works better for your purpose then I suggest you just use that instead...

Michael Vinther

> software developer <

10 (edited by bmibonn 2007-07-12 5:36:40)

Re: Is the Image Analyzer suitable for 3D-interferometry ?

Yes, the frequency cut off of profft must be sharp and was manually selected. As you could see very fine and sharp details  after filtering without any smoothing effects and any artefacts as a side ratio of 1:2.

Best regards
Weischer

11

Re: Is the Image Analyzer suitable for 3D-interferometry ?

Actually a sharp frequency cut off has nothing to do with a sharp image - rather it can be the other way around. A sharp edge (like the rectange you draw) in the frequency domain does not give sharp edges in the spatial domain - rather it gives "ringing" in the spatial domain as you see.

Michael Vinther

> software developer <

12 (edited by bmibonn 2007-07-15 9:11:42)

Re: Is the Image Analyzer suitable for 3D-interferometry ?

The FFT bandpass of PROFFT.exe has no rectangle but a circle. The results of the ProFFT bandpass FFT are excellent in comparison to the results of Image Analyzer.
The artefacts produced by Image Analyzer depending on the side ratio of 1:2, as demonstrated, cannot be accepted.
I think the FFT in Image Analyzer is not suitable for reducing the fringes of an interferometry image.

Best regards
Weischer

13

Re: Is the Image Analyzer suitable for 3D-interferometry ?

Unfortunately I have no plans for improving the FFT filtering in IA in the near future as the other filters in the software fulfills most needs.

Michael Vinther

> software developer <