matlab代写-BP5505:

时间：2021-03-03

BP5505: Hands-on mini-project - image processing tasks
Your task is to (a) download the files (images) from the links in the MyAberdeen page (the image names correspond
to the problem numbers shown below), then (b) decide, using the information given below and by examining each
image, what is 'wrong' with each one. You should then (c) try to solve each problem and (d) document for each one,
exactly what you think the problem was, and what you did to correct it. You should combine these documented
solutions into a report and submit this to MyAberdeen in the usual way, so that it can be checked by TurnitIn.
However, you should also submit your results to me by email, as a zip file attachment containing your report (which
you should name .docx), your scripts as *.m files, and your corrected images. This is because I need to
see your processing scripts also, not just your final, ‘improved or repaired’ images. You should name your scripts
solution01.m, solution02.m and so on, and your corrected images as corrected01.png corrected02.png etc. You
should name your zip file as .zip, so that I can still mark them anonymously.
Bear in mind that there may not be a 'correct' or 'perfect' solution to all of these images. What I will be looking for is
your approach to the problem: 'I thought the problem was ... I tried this ... and that ... and because of ... I found that
this approach ... was useful, but this other approach ... didn't help - I think that might have been because ...'.
Basically, tell us what you did, and why, and try to convince us you were trying reasonable things.
PROBLEM 1: This is a colour fundus (retina) camera image. Can you cheer up this patient?
They are undoubtedly suffering from the blues...
Full colour images are usually stored as a multi-band image consisting of the three colours:
red, green and blue. Since retinal images usually appear red, perhaps the user has confused
the ordering of the colour bands?
[ Difficulty: 1/5]

PROBLEM 2: Here is a Nuclear Medicine SPECT brain reconstruction study. It seems that
this patient is feeling a little prickly? What can be done to improve this image? Make sure
you adjust any filter parameters carefully!
[ Difficulty: 1/5]

PROBLEM 3: This is an image from the Colour Scanning Laser Ophthalmoscope (SLO). The
dark lines are certainly not correct. There are a number of approaches you could take to fix
this - suggest some, together with their advantages and disadvantages and choose one of
these for your own solution.
[ Difficulty: 3/5]
PROBLEM 4: Another image from the Scanning Laser Ophthalmoscope. The SLO acquires its
images in an interlaced fashion (i.e. all odd lines followed by all even lines). You will need to
look very closely at the image, and make sure you are viewing it at full resolution (e.g. if
using Xnview, set the zoom/magnification to 1:1) in order to see the problem. What might
have caused the artefact you can see in the image? What could be done about it? Hint: One
approach to fixing it could give you two different output images from this single image -
make sure you keep the correct aspect ratio.
[ Difficulty: 4/5]
COLOUR FUNDUS (RETINA)
CAMERA IMAGE

SPECT BRAIN
RECONSTRUCTION
COLOUR SCANNING LASER
OPHTHALMOSCOPE (SLO)
IMAGE

COLOUR SCANNING LASER
OPHTHALMOSCOPE (SLO)
IMAGE

PROBLEM 5: This image shows the typical speckle structure of diagnostic ultrasound. Centre
right is a kidney and there are some blood vessels at the left. See what you can do to "clean
up" this image so that the kidney with its internal structure is clearer. (Be warned that
ultrasound practitioners may still prefer the original, ‘uncleaned’ version!)
[ Difficulty: 2/5]

PROBLEM 6: This is an image of yeast cells (stained with Viablue-2), taken by attaching a
colour video camera to a fluorescence microscope. The stained cells are irradiated with
ultra-violet and fluoresce in the visible. Automated cell counting systems are interested in
the numbers of (a) clumps and (b) individual cells. Choose segmentation techniques (e.g.
smoothing for noise reduction? thresholding? other filtering or segmentation algorithms?)
so that you end up with two different ‘blob’ images optimised for (a) and (b) respectively.
You could then combine these again to obtain a single image showing both clumps and
individual cells. Perhaps you could even automatically count both the cells and the clumps?
[ Difficulty: 5/5]
PROBLEM 7: I'm not sure where this image came from or what its format is (it's certainly
not a tiff or a jpeg). I tried to open it by importing it into ImageJ as a ‘raw image’ – and it
came up looking a little strange, as you can see...
The small image here was my first attempt to open it. I've made a guess at the dimensions
and the pixel format - it's obviously not in the right position horizontally, but in addition,
the greyscales look a little strange. My guesses about the raw format must not be quite
correct; how can I open it correctly?
[ Difficulty: 2/5]

PROBLEM 8: Here is a photographic image – bit it has a ‘spotty’ appearance typical of some
printing processes. Can you help restore it to a less spotty state? (Hint: you may have seen this
image before, during your lectures …)
[ Difficulty: 3/5]

PROBLEM 9: There is a secret message hidden in this image. Can you read it and tell us
what it says? (Hint: you should perhaps examine the grey levels in the image very closely!)
[ Difficulty: 1/5]
PRINTED
IMAGE
ULTRASOUND
SPECKLE
STAINED
YEAST CELLS
CT CHEST
IMAGE
MRI SAGITTAL
IMAGE