Phase unwrapping, Wavelet transform profilometry (WTP), and MEtrology GUided RAdioTHerapy (MEGURATH) - PowerPoint PPT Presentation

1 / 39
About This Presentation
Title:

Phase unwrapping, Wavelet transform profilometry (WTP), and MEtrology GUided RAdioTHerapy (MEGURATH)

Description:

Phase unwrapping, Wavelet transform profilometry (WTP), and MEtrology GUided RAdioTHerapy (MEGURATH) Munther Gdeisat Phase unwrapping Phase unwrapping Problem ... – PowerPoint PPT presentation

Number of Views:245
Avg rating:3.0/5.0
Slides: 40
Provided by: munt2
Category:

less

Transcript and Presenter's Notes

Title: Phase unwrapping, Wavelet transform profilometry (WTP), and MEtrology GUided RAdioTHerapy (MEGURATH)


1
Phase unwrapping, Wavelet transform
profilometry (WTP), andMEtrology GUided
RAdioTHerapy(MEGURATH)
  • Munther Gdeisat

2
Phase unwrapping
3
Phase unwrapping
  • Problem statement
  • 2D phase unwrapping
  • 3D phase unwrapping
  • Publicityfree source code

4
What is phase unwrapping?
  • Suppose that you have a signal, an image or a
    video.
  • The required information is embedded in the phase
    of the signal, image or video
  • To extract the phase information you need an
    algorithm
  • Fourier transform
  • Wavelet transform
  • Etc
  • Normally, these algorithms produce complex data.
  • The phase can be extracted using the arctangent
    function that operates on these complex data.

5
What is phase unwrapping?
  • The arctangent function extracts the required
    phase information.
  • But this information cannot exceed the range -?
    and ?.
  • If the original information exceeds the 2? range,
    then
  • The extracted phase is wrapped
  • And an unwrapping algorithm is required.

6
1D phase unwrapping- example 1
  • Phase unwrapping appears simple at the first
    glance but in reality it is a very difficult
    problem.

7
1D phase unwrapping
8
1D phase unwrapping
9
2D phase unwrapping-example 2
Unwrapping using basic algorithm
An image
Unwrapping using our algorithm
The image is wrapped
10
Unwrapping using our algorithm
Unwrapping using LP algorithm
Unwrapping using Flynn algorithm
The image
11
Example 3
  • Doris Mannequin
  • Information encoded in the phase of the fringe
    pattern
  • The phase information are extracted using the
    Fourier transform. The resultant image is wrapped
  • Unwrapping the image to get the required height
    information.

12
2D phase unwrapping
  • Hundreds of algorithms already proposed in the
    literature
  • The robust phase unwrapping techniques are time
    consuming (10 seconds or more to process an image
    with the size of 512 ? 512 pixels)
  • Our algorithm is robust and requires 0.25 seconds
    to process an image with the size of 512 ? 512
    pixels
  • It is easy to understand-but hard to program
  • The C language source code is available to
    download for free from GERI website.

13
How the algorithm does work?
  • The quality of each pixel is calculated using its
    8 direct neighbours
  • Horizontal and vertical edges are constructed.
  • Edges connect two pixels together.
  • The quality of the edges are calculated
  • Search for the edge with the highest quality
  • Unwrap the two pixels that it connects or
  • Unwrap a pixel and a region that it connects or
  • Unwrap the two regions that it connects
  • Search for the next unprocessed edge with the
    highest quality
  • Unwrap the two..
  • And so on for the rest of edges

14
(No Transcript)
15
3D phase unwrapping
  • Sometimes we are required to unwrap a sequence of
    images (i.e., video data)
  • The existing algorithms are very time consuming
  • The famous prelude 3D phase unwrapper requires
    five days to process a volume of 512 ? 512 ? 50
    voxels.
  • Other unwrappers require few minutes to unwrap
    such data volume but they are not robust.
  • Our 3D phase unwrapper is fast, robust and free
    to download
  • It requires less than a minute to process this
    data volume.
  • Our 3D phase unwrapper is more robust than any
    other existing 3D phase unwrappers.
  • We have published three journal papers and two
    conference papers about 2D and 3D phase
    unwrapping.
  • One of these journal papers is very famous and it
    has been cited 19 times.

16
Phase unwrapping
  • We got many customers to our phase unwrappers
  • 1. The Christie Hospital, Manchester, UK.
    http//www.christie.nhs.uk/
  • 2. EADS Astrium, Germany. This is a multinational
    aerospace manufacturer of satellite systems. Its
    website is http//www.astrium.eads.net/
  • 3. The European Southern Observatory based in
    Australia. Their website is http//www.eso.org/pub
    lic/
  • 4. The Helen Wills Neuroscience Institute at the
    University of California, Berkeley, Their website
    is http//neuroscience.berkeley.edu/.
  • 5. Our phase unwrappers have been integrated in
    SciPy free open-source mathematical library. The
    website for the library is http//www.scipy.org/.
  • 6.Microgravity Research Centre  Universite Libre
    de Bruxelles
  • And many we do not know!!!

17
Phase unwrapping
  • Just search in Google the words phase unwrapping.
    This search will produce around 100,000 pages.
  • Search for the phrase phase unwrapping This
    will produce around 50,000 pages approximately.
  • We are number four after in both searches
  • Wikipedia
  • Stanford university (just a tutorial)
  • An Italian university (just a tutorial)

18
Wavelet transform profilometry
19
Wavelet transform profilometry
  • Wavelet transform profilometry means the
    utilisation of the continuous wavelet transform
    to extract the phase information from a fringe
    pattern.
  • There are two methods to perform this task
  • One-dimensional continuous wavelet transform
  • Already exists in the literature
  • Two-dimensional continuous wavelet transform
  • Has been developed in GERI

20
Wavelet transform profilometry
  • This algorithm has many advantages over the
    Fourier transform method
  • little edge distortion
  • The extracted phase is more accurate
  • No limit on the number of fringes in an image
  • Better noise performance

21
Example 1 a fringe analysed using wavelet and
Fourier methods
Fringe pattern
22
Wrapped phase maps using
Fourier transform
Wavelet transform
Fourier transform
Wavelet transform
Fourier transform
23
Unwrapped phase map using
Wavelet transform
Fourier transform
24
A fringe pattern with two fringes only analysed
using the wavelet method
25
The wavelet method can deal with holes in a
fringe pattern image
26
Wavelet software
  • We (Munther and Abdul-Basit) have developed a
    software in C that is capable of extracting the
    required phase information from a fringe pattern
    using the wavelet method.
  • I will publish this software on GERI website and
    make it available to download for free
  • Repeating the phase unwrapping good experience

27
Wavelet software
  • We have spent over one year in writing this
    wavelet software.
  • The execution time required to analyse a 512 X
    512 pixels fringe pattern in Matlab using the
    wavelet method is approximately five minutes.
  • The same operation requires less than one second
    using our software.
  • We have been invited to write a tutorial paper
    about this topic to the Optics and Lasers in
    Engineering Journal.
  • We have published three journal papers and three
    conference papers about this topic

28
MEtrology GUided RAdioTHerapy(MEGURATH)
29
Headlines
  • Introduction
  • Specifications
  • Components
  • Processing Pipeline
  • Future Work

30
MEGURATH
  • This system is designed and built to enable the
    medics to measure the height map of human bodies.
  • This is an important step in cancer radiotherapy
    treatment

31
Introduction
  • A state-of-the-art 3D body measurement system
    using fringe analysis
  • A catalogue of fringes analysis, phase unwrapping
    and fringes calibration methods were employed
  • User preferences of speed and robustness
  • System currently uses only one sensor unit (CCD
    camera and LCOS projector)

32
3D measurement System Block Diagram
  • Monochrome cameras
  • Optical filters to get three channels
  • XB08 (Blue)
  • XB21 (Green)
  • XB24 (Red)

33
Specifications
  • Field-of-view 40cm x 40cm x 40cm
  • Measurement speeds
  • From 5 to 14 fps depending on pre-adjusted
    settings
  • Phase stepping
  • Fourier
  • Wavelet
  • Maximum error 0.5 mm

34
Components
1/4
  • System simply consists of projector, camera,
    processing station and processing software

35
Components
2/4
  • Projector Canon XEED SX60
  • LCoS technology
  • Advantages
  • Low pixelisation
  • High brightness (2500 lumens)

36
Components
3/4
  • Camera Prosilica GE1380
  • GigE technology
  • Monochrome
  • Progressive scan CCD
  • Sensitive Camera
  • 20 fps _at_ 1360 1024
  • 35 fps _at_ 512 x 512
  • Direct image registration to the system memory
    via a compatible Gigabit port

37
Components
4/4
  • Processing software

38
Processing Pipeline
  • Multithreaded processing framework
  • Input Thread Project and grab frames
  • Processing thread Apply measurement, unwrapping
    and calibration techniques
  • Output thread Save/display results

39
Thank you for listening.Any questions or
suggestions?
Write a Comment
User Comments (0)
About PowerShow.com