5th Intensive Course on Soil Micromorphology

1
5th Intensive Course on Soil Micromorphology
Naples 2001
12th - 14th September Image Analysis
Lecture 2 Image Acquisition Historic Aspects
2
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
The four Is of Microfabric Analysis

• Image acquisition
• normal photographs (via scanner)
• optical microscope (via digital camera)
• electron microscope (directly via A/D converter)

• Image Processing
• Image enhancement
• Filtering
• Image reconstruction
• Image Analysis
• Quantitative analysis
• Interpretation

3
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition

• Image Acquisition - Requirements
• Spatial Resolution
• Intensity resolution
• Illumination issues
• Image distortion
• Image formats

4
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Image Acquisition - Requirements Spatial
Resolution Photographic Recording Photographic
enlargement is possible often up to 10 times
depending on grade of film. Low magnification
covers larger area, and detail can be seen with
enlargement. Digital Recording NOT possible to
enlarge to see detail. Magnification /
resolution must be selected at outset. May be a
compromise between area covered and detail.
5
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Spatial Resolution- Digital Recording Image
divided into pixels which are the smallest
element that can be resolved. Area covered by
each pixel depends on the field of view and this
will vary with the magnification in use. Field of
view - 0.1 mm ( x1000 in the SEM). Digital
recording medium resolution - 512 x 512
pixels gtgtgtgt each pixel will represent about 0.2
?m.
Enlargement of (a) by x 4 gives no additional
information. Isolated features (c) 1 pixel wide
can be resolved, but not if they touch (e).
Pixel size should be no more than 50 of
smallest feature.
6
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition

• Most modern frame stores /cameras have a minimum
  resolution of around 400 - 500 pixels
• Ideally, for high quality analysis 1000 - 2000
  pixel device should be used.
• One SEM manufacturer has an option of a frame
  store gt 3000 pixels.

Spatial Resolution- Digital Recording
In the SEM double number of pixels gtgtgt 50
reduction in magnification
increases the area
covered by a factor of 4. but may not be
entirely true if beam diameter is adjusted with
magnification
7
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition

• Scanners
• The resolution on these may be
  adjusted
• Often set at default values
  between 50 and 200 dpi
• options often up to 1200 dpi

Spatial Resolution- Digital Recording
For a 150 x 150 mm image and 100 dpi, image is
600 x 600 pixels. Each pixel ? 0.25 mm At 1200
dpi, the image will be very large at 7200 x 7200
pixels Each pixel ? 0.02mm For a black and
white image (grey-level image) storage 100 dpi
gtgt 0.36Mbyte storage (for 150 x 150 mm image)
1200 dpi gtgt 50 Mbyte for a single image.
8
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Intensity resolution
Human eye can resolve about 16 grey levels at
best Monochrome publications can achieve 8 at
best. Digital Images Binary image Black and
White only pores black - solids white
or vice versa Grey Level Image 0 - 15 grey
levels 0 - 255 grey levels 0 - 65535 or
higher False Colour 0 - 255 pseudo colours True
Colour 16 million colour shades
9
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Intensity resolution
Binary images Binary images are basis of much
image processing e.g. particle/feature size
analysis. Each pixel is either 1 (foreground) or
0 background and may be stored in a single
bit. Data storage is usually in form of BYTES -
or 8 bits. 8 pixels stored in each byte. (32
kbytes needed to store image. Most images are not
in binary form and must be converted before most
image processing/analysis packages can be used.
10
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Intensity resolution
Storage values for 512 x 512 image
Grey Level Images 16 grey levels (used by older
devices). store data from 2 pixels in each byte
gtgt128 kBytes BYTE 256 grey levels most
common gtgt 256 kBytes range 0 -
255 must be integer values INTEGER 65536 grey
levels 2 bytes per pixel gtgt 512 kBytes
range -32376 - 32375 must be integer
values REAL Higher grey level resolutions
REAL format storing as 4 bytes per pixel
gtgt 1024 kBytes minimum
intensity -1 x 1038 maximum 1 x 1038
11
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Intensity resolution
False Colour Usually equivalent to 256 grey
level images individual grey level shades are
shown by colour shades to overcome limitations of
human eye. True Colour Often as RGB (red green
blue) - most common each colour has an
intensity range 0 - 255 giving 16 million
possible colours. 3 bytes (24 bit) are needed
for each pixel Sometimes CYM (cyan yellow
magenta) is used Alternatively HSI (Hue
Saturation Intensity)
12
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Illumination issues

• Binary Images often needed for analysis
• BUT
• Illumination Settings at acquisition can often
  bias the results
• brightness and contrast of illumination gtgtgt
  saturation
• brightness and contrast setting on SEM gtgtgt
  saturation
• intermediate photographic developing and printing
  
• 
  gtgtgt saturation

May want high contrast (and saturation for binary
images) BUT in other cases full dynamic range
should be retained.
13
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Illumination issues
Though (b) or (c) may look better, (a) contains
more information
14
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Illumination issues
full 0 - 255 range
Over-saturated
Some image acquisitions systems when using byte
(0 - 255) range - reflect range when saturation
occurs. 256 gtgtgt 0, 257 gtgtgt 1 etc. - see effect
in bright areas
15
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Illumination issues
Non-uniform illumination will create problems in
later thresholding and analysis
16
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Image Distortion
Optics of microscope can distort image gtgtgt
problems with measurement
17
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Image Distortion
Microscope should be calibrated with a grid
18
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Image Formats
BMP TIFF JPG GIF ------Widespread Some
formats require key information to be stored in
another file problems if associated file is
removed Most formats have key header information
in file itself. Most compact formats use BYTE (0
- 255) format can be swapped easily between
platforms BUT INTEGER format (2 bytes) may be
problematic - swapping between PC/Macintosh/UNIX
(as high and low bytes are swapped). Even more
problems if REAL/ COMPLEX format is used
19
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Image
acquisition
Image Formats
Typical Image 512 x 512 x 8 gtgtgtgt 256 kBytes True
Colour Image 4096 x 4096 x (24bit) would occupy
more than 50 Mbytes Special Formats - (RLE) for
Binary Images and selected Classified Images
arising from Image Analysis - more efficiently
stored in a Run-Length-Encoded format
(RLE). Applicable for BYTE and INTEGER format if
large areas have the same intensity value -
compression up to 90 is possible. May need to
re-expand image back to normal BYTE or INTEGER
format before using for Analysis purposes. Cannot
be used on grey-level images
20
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Historic
Aspects
Historic Development of Image Quantification
Optical Methods
Optical Diffraction was used as a method to
assess orientation in 1970s. Related technique
of Convolution Square was also used
21
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Historic
Aspects
Historic Development of Image Quantification
Upturned plates on sand grain The diffraction
pattern indicates preferred orientation
direction, degree of orientation and spacing of
features
22
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Historic
Aspects
Historic Development of Image Quantification
A high degree of orientation in consolidated clay
shown by shape of diffraction pattern. Note
inverse spacing relationship between diffraction
pattern and image
23
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Historic
Aspects
Digital equivalent of Optical Diffraction
The Optical Diffraction Pattern of images may be
computed digitally as Fourier Transform - this
is exploited in image reconstruction in Lecture 10
24
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Historic
Aspects
Digital equivalent of Convolution
Note unlike diffraction the spacing of features
in convolution is directly related to spacings in
image
25
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Historic
Aspects
Stereoscopy/Photogrammetry
SEM is ideal for photogrammetric measurement
using stereo-photographs Geometry of SEM should
be considered with viewer at the electron source
and illumination at the electron collector - i.e.
and apparent reversal of convention.
26
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Historic
Aspects
Geometry of SEM
At low magnification lt 500 central
projection At high magnification gt 1000x parallel
projection
27
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Historic
Aspects
Unconsolidated Kaolin - picture width 6.2 ?m
28
5th Intensive Course on Soil Micromorphology -
Naples 2001 Image Analysis - Lecture 2 Historic
Aspects
Stereo-photogrammetry
Stereo photograph and associated Pole Diagram
showing preferred 3 - D orientation