Talk:Gabor filter

	This article is rated Start-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects:
Mathematics Low‑priority

Gabor wavelets have many applications in biologically inspired systems, especially those related to computer vision, since their responses are similar to those of the nerves in the V1 region in the visual cortex.

I know very little about Gabor filters, but when I tried implementing one I ran into problems with sigma. After some poking around I found this page (http://matlabserver.cs.rug.nl/edgedetectionweb/web/edgedetection_params.html) that says ${\displaystyle \sigma }$ must be specified through b. Could someone w/ more knowledge on the subject please comment? —Preceding unsigned comment added by Jason.surratt (talk • contribs) 15:03, 25 November 2008 (UTC)[reply]

I believe the Movellan reference has been updated to: http://mplab.ucsd.edu/tutorials/gabor.pdf —Ebrevdo (talk) 08:15, 20 January 2009 (UTC)[reply]

I am really surprised by the definition of the 2d Gabor filters. I work in the field of image processing, where Gabor filters are regularly used and I have never seen a definition like this one. This is only one part of the filter - the real part. Gabor filters must be complex or they do not make any sense. Only then they exhibit some of the characteristic properties, such as optimal localization in the spatial and frequency domains. —Preceding unsigned comment added by 193.95.243.51 (talk) 07:50, 26 July 2010 (UTC)[reply]

I second that, we need the full definition. As I understand it, for most vision applications what you take in the end is the magnitude of the complex vector defined by the responses of the real and imaginary part of the filter response. Additionally when it comes to implementing gabor wavelets it is very important to use the version with the DC component removed from the cos part.

I second that as well. Seeing that the introduction speaks about "edge detection", and the article mentions only the even filter which is actually a line detector rather than an edge detector, the full filter pair definitely needs to be described. The presentation is confused by the fact that the image shown is an odd (sine component) kernel, but the equation presents an even (cosine) kernel.

This is the wrong definition and should relate to the work of Jones and Palmer (1987) who showed that the real part of complex Gabor functions fit very well the receptive field weight functions found in simple cells in a cat's striate cortex. I will correct the page later. —Preceding unsigned comment added by Mikeblew (talk • contribs) 13:38, 24 August 2010 (UTC)[reply]

"a 2D Gabor filter is a Gaussian kernel function modulated by a sinusoidal plane wave" "the Morlet wavelet, ... a constant κσ subtracted from a plane wave and then localised by a Gaussian window"

So in the 1D case, are Gabor and Morlet wavelets the same thing? — Preceding unsigned comment added by 71.167.62.119 (talk) 15:52, 3 June 2011 (UTC)[reply]

The matlab code is probably very informative if you already know exactly what a Gabor filter does. If you don't, this unnecessarily complicated and uncommented piece of code might as well have been left out completely. Could somebody add a comment to each line describing its use? For instance, why ceil over max(1,xmax), you might as well take 1 as a default and forego the entire definition of xmax. — Preceding unsigned comment added by 145.18.248.164 (talk) 16:01, 6 November 2012 (UTC)[reply]

Seconded. The large section of code should be accompanied by descriptions or comments to explain functional flow in order to help both those who are familiar with MATLAB, as well as those who aren't. Otherwise the presence of a large block of code, despite its relevance, is unnecessary. ThsTorturedSoul (talk) 06:25, 18 May 2014 (UTC)[reply]

The definition given here appears to be from reference 2, the masters thesis of Jesper Juul Henriksen. The formulas are in Appendix B. Cross comparing references, it appears that at minimum Henriksen dropped a term in equation B.2 for the imaginary part. The wikipedia entry is missing this term for both the real and imaginary parts. I have marked the definition page as a disputed section so nobody else wastes half a day on suspect formulations. Ckelloug (talk) 18:55, 28 February 2013 (UTC)[reply]

There's a nice picture (captioned 'Example of a two-dimensional Gabor filter'), which would probably go a long ways towards explaining this to those of us who want the intuition, but not necessarily the math. Unfortunately, afaict the picture is not referenced or explained in the text, and it's not self-explanatory. E.g. why are there two deeply colored areas, with different colors? What do the colors mean? Is there an intuition of what this particular filter "captures", based on those colors? (e.g. it captures a line at 45 degrees, or...) Mcswell (talk) 23:03, 28 December 2016 (UTC)[reply]

Hello fellow Wikipedians,

I have just modified one external link on Gabor filter. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:

• Added archive https://web.archive.org/web/20090419123314/http://mplab.ucsd.edu:80/tutorials/gabor.pdf to http://mplab.ucsd.edu/tutorials/gabor.pdf

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

This message was posted before February 2018.

• If you have discovered URLs which were erroneously considered dead by the bot, you can report them with this tool.
• If you found an error with any archives or the URLs themselves, you can fix them with this tool.

Cheers.—InternetArchiveBot (Report bug) 01:25, 7 January 2017 (UTC)[reply]

The article, especially the introduction section is way too technical. It should be possible to explain the concept without referring to "Gaussian kernel function modulated by a sinusoidal plane wave" at the very beginning.

82.33.29.34 (talk) 00:58, 9 February 2017 (UTC)[reply]