de-focus aberration?

Share icon

Thu, 03/03/2016 - 10:09 am

Hi,

Attached is an image i'm trying to improve/correct.... The left image shows the collected image, and the right image shows the image with a threshold overlay. I think the threshold is suggesting there is a de-focus aberration in the optical system that could potentially be described by a Zernike Polynomial. Is there a way to use "ZernikeR(n,m,r) to model/fit and then subtract that aberration?

Thanks,

Clayton
Optical aberrations and the Zernike polynomials concern wave-front phase distortions. Your image provides only intensity information. Unless you can provide additional input to your problem, you will have a difficult, if not impossible, job of teasing out the missing phase information.

From the apparent symmetry of your thresholded image near the center, if I had to guess, I would suspect a weak stray vignetted reflection. Or is this the "aberrated" feature you wish to correct?

March 8, 2016 at 05:29 am - Permalink

You are really not providing enough information about your images. In a very abstract sense, given an image (intensity not a complex field) you have a non-unique inverse problem. If you know something about the "source" you might be able to use various methods for estimating the aberrations but as was pointed out above, the lack of phase information is going to make your task difficult. If you have reason to think that your system suffers from aberration X, you could test your hypothesis by constructing a corresponding PSF and using ImageRestore.

A.G.
WaveMetrics, Inc.

March 9, 2016 at 11:44 am - Permalink

Thanks for the details. Sounds like the topic is more complex than originally considered.

March 16, 2016 at 11:15 am - Permalink

While I have you reading this post A.G. I would appreciate to see the code which determines the "rectangularity" of objects found via ImageAnalyzeParticles. Is that code accessible and if so, where to locate?

Thanks

March 16, 2016 at 11:19 am - Permalink

ctmckee wrote:
While I have you reading this post A.G. I would appreciate to see the code which determines the "rectangularity" of objects found via ImageAnalyzeParticles. Is that code accessible and if so, where to locate?


The connection with the OP is not immediately obvious.

The documentation defines the rectangularity as a ratio of areas. Once you have the area of the particle, and the pixels that define the limits in the X and Y directions, you can compute the ratio. I think it sounds more complicated than it is.

A.G.
WaveMetrics, Inc.

March 16, 2016 at 01:30 pm - Permalink

Thanks, that would be simple enough to code.

I was confused, because the documentation says it is for an "inscribing" rectangle. I assumed that meant that all pixels of the rectangle had to be within the contour of the object. Your response clarifies that the the rectangle is defined by the limits of X and Y points of object; meaning for some objects, portions of the rectangle could lie "outside" the object? Is this a distinction between "inscribing" and "inscribed"?

March 18, 2016 at 09:36 am - Permalink

ctmckee wrote:
Thanks, that would be simple enough to code.

I was confused, because the documentation says it is for an "inscribing" rectangle. I assumed that meant that all pixels of the rectangle had to be within the contour of the object. Your response clarifies that the the rectangle is defined by the limits of X and Y points of object; meaning for some objects, portions of the rectangle could lie "outside" the object? Is this a distinction between "inscribing" and "inscribed"?


That is correct. An "inscribing" rectangle is one that contains the whole object, i.e., all pixels of the object lie inside the rectangle but the rectangle may contain pixels that are outside the object.

A.G.

March 18, 2016 at 12:05 pm - Permalink