Billedanalyse af AFM billeder

From Midtiby

(Difference between revisions)
 
(5 intermediate revisions not shown)
Line 14: Line 14:
# LSQ Legendre polynomial fitting, m.m. [http://www.tau.ac.il/~lab3/MATLAB/]
# LSQ Legendre polynomial fitting, m.m. [http://www.tau.ac.il/~lab3/MATLAB/]
-
= Billedanalyse tekst =
+
'''Referencer'''
-
 
+
# Particle Characterization with AFM (Natasha Starostina, Paul West 2006) [http://nanoparticles.pacificnano.com/nanoparticle-characterization.html]
-
[[Image:AfmTraditional.png Traditionel behandling af et AFM billede.]]
+
#: Care should be taken when processing the image to avoid introducing artificats into the image. A common artifact is shown in Figure 1B. The bands in the image are derived from line by line leveling without using the threshholding. Such bands can result in erroneous nanoparticle characterization.
-
[[Image:AfmPotentialOrder2.png Potential minimering]]
+
# AFM Image Artifacts (Paul West, Natalia Starostina) [http://www.lot-oriel.com/site/site_down/pn_artifacts_deen.pdf#search=%22afm%20tilt%20bow%22]
-
 
+
#: '''1''' The piezoelectric scanners that move the probe in an atomic force microscope typically move the probe in a curved motion over the surface. The curved motion results in a “Bow” in the AFM image. Also, a large planar background or “Tilt” can be observed if the probe/sample angle is not perpendicular.
-
== Problembeskrivelse ==
+
#: Often the images measured by the AFM include a background “Bow” and a background “Tilt” that are larger than the features of interest. In such cases the background must be subtracted from the image. This is often called “leveling” or “flattening” the image. After “leveling” the desired features are typically directly seen in the image.
-
 
+
#: '''2''' As mentioned in section 2.4, most images have some tilt and bow that is introduced to the images by the scanner or stage configuration. There are a number of background subtraction options that are possible. The two most common types are:
-
Benyttes en skanning probe teknik til at lave mikroskoperinger, skal
+
#:Line by line leveling - 0 to 4(th) order
-
de rå data efterfølgende behandles.
+
#:Plane Leveling        - 0 to 4(th) order
-
 
+
#:Also, software typically allows you to exclude areas from the leveling. When an area is excluded, it is not used for the calculation of the background in the image.
-
== Klargøring af billede (preprocessering) ==
+
# New method to estimate step heights in scanning-probe microscope images (Hal Edwards 1997) [http://www.iop.org.heimdal.bib.sdu.dk:2048/EJ/article/0957-4484/8/1/002/na7102.pdf]
-
 
+
#: Abstract. A new algorithm, the polynomial step-function fit (PSFF), is presented. The PSFF algorithm extracts step heights from noisy and distorted scanning-probe microscope (SPM) images. A one-dimensional, line-by-line implementation as well as a two-dimensional, full-image version are presented. The PSFF algorithm allows the correction of image distortions due to nonlinearities in the piezoelectric scanner and Abbe offset errors, but piezoelectric creep and hysteresis must be corrected separately, and may set the ultimate physical limitations on the accuracy of the PSFF algorithm. The PSFF algorithm is demonstrated with a real sample.
-
Ved at udnytte specifik viden om de optagne data (at det er målinger
+
# Influence of data analysis and other factors on the short-term stability of vertical scanning-probe microscope calibration measurements (Hal Edwards 1997)[http://scitation.aip.org.heimdal.bib.sdu.dk:2048/getpdf/servlet/GetPDFServlet?filetype=pdf&id=JVTBD9000016000002000633000001&idtype=cvips&prog=normal]
-
på to diskrete højder), forbedres den direkte databehandling.
+
-
 
+
-
'''Plot af en oprindelig linie (ret linie med meget afvigelser)'''
+
-
 
+
-
'''Plot af linie efter at den generelle tendens er fjernet'''
+
-
 
+
-
Det der ønskes er at baggrunden gøres så ensartet som muligt.  
+
-
Derfor er det interessant kun at fitte et polynomie til baggrundsniveauet
+
-
også fratrække dette.
+
-
 
+
-
=== Traditionel fremgangsmåde ===
+
-
 
+
-
Der fittes et polynomie af grad n (n = 1 eller 2) til en skanningslinie,
+
-
derefter trækkes polynomiet fra skanningslinien.
+
-
Denne operation fjerner fejl der skyldes overordnede fejl fra to kilder:
+
-
* at prøve og probe ikke er i samme plan
+
-
* at prøven bøjes
+
-
 
+
-
'''Problem'''
+
-
 
+
-
=== Skelne mellem de enkelte niveauer ===
+
-
 
+
-
Den traditionelle fremgangsmåde kan forbedres ved at skelne mellem de
+
-
to mulige niveauer, sådan at der kun fittes et polynomie til et af
+
-
niveauerne, f.eks. baggrunden.
+
-
 
+
-
Problemet er at skelne mellem niveauerne når der er så meget støj som
+
-
der er i de benyttede data.
+
-
 
+
-
=== Et optimeringsproblem ===
+
-
 
+
-
Opgaven med at gøre baggrunden ensartet, kan defineres som et
+
-
optimeringsproblem.
+
-
Det gøres ved at definere en fejlfunktion der har et globalt minimum
+
-
når baggrunden er fratrukket optimalt.
+
-
En sådan funktion kan kun opstilles når nogle af egenskaberne af
+
-
dataene er kendt.
+
-
I dette tilfælde vides det at de målte data svinger mellem to højde
+
-
niveauer med en fast afstand.
+
-
 
+
-
'''Eksempel på et godt signal med markeringer af de to faste niveauer'''
+
-
 
+
-
Fejlfunktionen f(A) er defineret ved
+
-
:f(A) = \\sum_{i = 1}^{length(A)} min(abs(A_i), abs(A_i - h))^{order}
+
-
hvor
+
-
*'''h''' er adskillelsen af de to niveauer
+
-
*'''order''' er en parameter der styrer hvilke afstande der
+
-
skal vægtes kraftigst.  
+
-
 
+
-
 
+
-
== Finde objekter ==
+
-
 
+
-
== Beskrive kanter af objekter ==
+

Current revision as of 20:24, 28 August 2006

Forskellige links jeg har samlet til et projekt omkring billedanalyse af AFM billeder.

Matlab og billedbehandling

  1. MATLAB and Octave Functions for Computer Vision and Image Processing [1]
    Guld side!!!

Ikke sorteret

  1. Image Feature extraction for Classification purposes [2]
  2. Mathtools: C++ Image Processing [3]
  3. Matlab tutorial [4]
  4. COURSES ON MATHEMATICAL MORPHOLOGY [5]
  5. Fundamentals of Image Processing [6]
  6. Erode og dilate uden IPT [7] [8]
  7. LSQ Legendre polynomial fitting, m.m. [9]

Referencer

  1. Particle Characterization with AFM (Natasha Starostina, Paul West 2006) [10]
    Care should be taken when processing the image to avoid introducing artificats into the image. A common artifact is shown in Figure 1B. The bands in the image are derived from line by line leveling without using the threshholding. Such bands can result in erroneous nanoparticle characterization.
  2. AFM Image Artifacts (Paul West, Natalia Starostina) [11]
    1 The piezoelectric scanners that move the probe in an atomic force microscope typically move the probe in a curved motion over the surface. The curved motion results in a “Bow” in the AFM image. Also, a large planar background or “Tilt” can be observed if the probe/sample angle is not perpendicular.
    Often the images measured by the AFM include a background “Bow” and a background “Tilt” that are larger than the features of interest. In such cases the background must be subtracted from the image. This is often called “leveling” or “flattening” the image. After “leveling” the desired features are typically directly seen in the image.
    2 As mentioned in section 2.4, most images have some tilt and bow that is introduced to the images by the scanner or stage configuration. There are a number of background subtraction options that are possible. The two most common types are:
    Line by line leveling - 0 to 4(th) order
    Plane Leveling - 0 to 4(th) order
    Also, software typically allows you to exclude areas from the leveling. When an area is excluded, it is not used for the calculation of the background in the image.
  3. New method to estimate step heights in scanning-probe microscope images (Hal Edwards 1997) [12]
    Abstract. A new algorithm, the polynomial step-function fit (PSFF), is presented. The PSFF algorithm extracts step heights from noisy and distorted scanning-probe microscope (SPM) images. A one-dimensional, line-by-line implementation as well as a two-dimensional, full-image version are presented. The PSFF algorithm allows the correction of image distortions due to nonlinearities in the piezoelectric scanner and Abbe offset errors, but piezoelectric creep and hysteresis must be corrected separately, and may set the ultimate physical limitations on the accuracy of the PSFF algorithm. The PSFF algorithm is demonstrated with a real sample.
  4. Influence of data analysis and other factors on the short-term stability of vertical scanning-probe microscope calibration measurements (Hal Edwards 1997)[13]
Retrieved from "http://editthis.info/Midtiby/Billedanalyse_af_AFM_billeder"
Personal tools