Biomedical Engineering Reference
In-Depth Information
before images are recorded. However, under some circumstances it may be desirable to
change the scale of an AFM image. For example, where an internal size standard in the
image is present, and the measured dimensions are not accurate, this can be a way of
correcting the calibration in the image. If images of the same feature with two different
instruments are to be directly compared, it may be useful to rescale them so that they show
features as exactly the same size. If the AFM is recalibrated during a set of experiments, it
can also be useful to change the scale of the earlier images to match the known calibration.
5.1.4 Error correction
AFM images often contain 'errors', which typically result from unwanted interactions of
the tip with the sample, such as sample movement under the tip, or strong tip-sample
forces leading to vibration or streaking in the images. They can also be the result of
external forces such as vibration or acoustic noise. If it's not possible to acquire a new
image without the errors, the user may wish to remove them. This can be achieved by
specific routines in the AFM processing software. Line removal can be done by removing
a single line and replacing it with the average of the two lines next to it. A 'glitch' may be
removed by replacing the glitched pixel with the average of the eight pixels around the
unwanted pixel. These sorts of corrections should be used with great caution, and images
corrected in this way should not be used for further analysis, as the correction process will
change the data. For example, data that was subjected to error correction will exhibit lower
roughness values after treatment, and these values do not reflect the true roughness of the
surface. A further type of image correction commonly applied to AFM data is deconvolu-
tion. This process attempts to remove the effect of the finite width of the AFM tip (i.e.
tip-sample convolution) from the AFM data. Almost all AFM processing software has
routines to carry out deconvolution. This is best carried out when the shape of the tip is
well known (either by blind estimation or measurement). Blind estimation of the tip shape
is often also included in the software. The use of these procedures is described in greater
detail in Section 2.5.
A large amount of information can be gleaned from AFM images just by viewing them.
However, the information visible in an image depends on how it is displayed. Furthermore,
the AFM user never works in isolation, and needs to communicate the information present
in the images to others. While much of this is done by the analysis techniques discussed in
the next section, the ability to display images that show what we want is vital, to make use
of AFM data. Often particular features can be enhanced and seen much more easily by
changing display parameters. Although there are many different display functions in AFM
data analysis software, in general the display functions don't alter the data in any way.
5.2.1 Histogram adjust
An image histogram is a plot of the height (or other parameter recorded in the
z
scale) of
each pixel, versus frequency. The most common way of displaying AFM data is as an
image where colour represents height. The AFM software will normally stretch the colour
