Tomographic image reconstruction using LabVIEW acquired data

[ZJ.Gu]

Hi mishklar,

Could you please tell more specific info about "impossible to debug"? I don't think the performance would be very poor in LabVIEW as 3D filterd backprojection is really filtering operation.

Thanks!

[mishklyar]

Hi ZhijunGu,

I mean, while developing some algorithm, I usually want to track every stage, look at variables, at data.There are aslo some constants, that becomes wires

at labview.Nothing bad regarding the performance, labview is compiled language,and very user friendly for multy-processor developement.But, say you have

5 constants, 4 arrays, 2  matrixes and one 3D matrix. All this inside 3 loops, so you have an indexes. So, at labview you have 12 wires, some of them looks the same,indexes also looks the same.

   At matlab you have 12 variables with different names.Personally, I prefer "indexX,indexY,indexZ" over three wires with a blue color  🙂

Michael.

[E.Garcia]

Hi MIchael.

It sounds to me that in similar situations you could benefit from using a custom probe in LabVIEW to debug your code.

A custom probe will allow you to watch numeric values, as well as combine them with other control representations such as graphs and even image controls.

Much flexible for debugging.

Ernesto

[Curt_Corum]

Here is a document with filtered backpriojection code.

It uses simulated projection data but you can modify to load data from a file.

http://decibel.ni.com/content/docs/DOC-15351

[ZJ.Gu]

Hi CurtCorum,

Thanks for your contribution! That is really useful for LabVIEW users to learn how to write backprojection with or without filtering.

ZJ Gu

[pajogdand]

Respected Sir,

          I am trying to apply wavelet trnasform for signal obtained after radon transform .But cant get better resolution Ican send me ur filter desinged or paper that u have refer for designing this filter.

[Curt_Corum]

Dear Pajogdand,

Not sure if you have resolved this isssue.

Backprjection filters and the related topic of sample density correcton are a very rich topic. Are you concerned with smothing resolution effects, sampling artifacts which may show up in a wavelet transform decomposition?

Basic information on types of filters for backprojection can be found it:

Foundations of medical imaging

http://books.google.com/books?id=UWhrAAAAMAAJ

Here is a reference (by no means the only) on the topic:

Bandlimited and Haar filtered back-projection reconstructions

http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=310874

Hope this helps,

Curt

[xbash]

Dear All,

I'm not sure this is the correct place to post my question, instead of image reconstruction. I'm trying to apply radon transform to an image. I think I got the main part correct but there are artifacts (pic below) appearing on the sinogram. And those artifacts always occurs at 45, 135,225,315 degree.(diagonal theta)

I think is due to the resolution of the pixels at those angle where the algorithm could not cope. I have attached the vi( it take a minute or so to run ) as it's not the most efficient algorithm.

Is there a way to get around the artifacts ? 

Regards

Tze

[Curt_Corum]

Tze,

Definatly it is possible, I did not look at your vi, but you have a bug in your code or algorithm.

You must have something that happens at 45 deg. in the very worst case if you cant find the problem, you can treat 45 deg, etc as special case and just add up without interpolation or whatever you are doing. (Trace along each all diagonals, not just main) but better to find the error.

Curt

[xbash]

Thanks Curt,

Looking into it further, at 45 degree and etc , when the "rho" is moving from one end to another end  it integrate a thick lines and thin lines and alternate between those hence the dark and bright pixels alternates. In other words at 45 degree the algorithm could not distinguish the delta function. At the moment I can't see how I can distinguish the delta value accurately

Please advice

Tze