Forums
Bruker Media
Community Media
Bruker AFM Probes
SPM Digest
Application Notes
NanoTheater
Website
中文
Brochures & Datasheets
Publications
Probes Catalog
Events
Manuals & Documentation
Presentations
Guide to AFM Modes
News
Journal Club
Webinars & Video
Nanovations
Other
>> I am a user with a Dimension 3100 AFM with a Nanoscope 4 controller.I have a controller software 5.31r1 version. >> I noticed that in modify menu there is a detrend function. I want to know what operation did it work? is that a dilation for tip?
I can show two papers talking about the popular methods to detrend the data.one is FFT base conventional , FFT the data and deconvolute with instrumental response function and inverse FFT. another one is nonlinear erosion list in my attachment by John Villarrubiain V102 N4 July 1997 J.Nist.I have no idea what methods veeco are using to detrend data. I read the manual there mentions the eigenvalue method. I am interesting in this because different detrend method requires different sequence to apply filters. Hereby I want to ask what if Veeco suggested sequence to detrend data?1. measure data from AFM?2. flatten?3.detrend?4.filter LOW/high pass?5. PSD?
Thanks for your help!
Chen,
I am a software engineer at Veeco Instruments and would like to assist you with your recent question about the Detrend analysis in v5.31r1.
First, I should clarify that this is not a tip dilation or deconvolution analysis. In fact there is no tip deconvolution analysis in our Instrumentation software although you could perhaps use the rolling ball filter for very simplified models of certain tips. In your email you referenced Viallarubia's work. We have discussed Villarubia's work with him in the past and even talked about joint projects but it hasn't resulted in an analysis so far.
Secondly, what Detrend does is to use PCA (principal component analysis) to find the primary eigenvectors and eigenvalues of the variance matrix and reconstruct the image from the first few eigenvectors and eigenvalues, thus smoothing and compressing the image. I'm sure you already understand this but a good refresher to this is at http://www.uwlax.edu/faculty/will/svd/index.html. So you can see that the number of eigenvalues determines the amount of smoothing. I agree with you that the reference manual needs to be updated to provide more information about Detrend.
Finally, you asked about a suggested sequence to detrend data and mentioned Flatten, Detrend, Filters etc. This is a difficult question since it's impossible to offer a general solution that will work for every image or application. We would need to see some sample images and get a better idea of your application for these images and then we could provide you with a recommended sequence of methods to use in your Nanoscope software.
I hope this has helped answer your questions. Please feel free to contact me if you need more information.
Regards,
Rohit Jain