Update Presentation 3

Image Processing

Peter Griffiths and Ethan Hilton

02 00 PM ,Wed, Oct 28 2015

In our last presentation… pic1.1

pic1.1

pic1.1

pic1.1

About the Data

5 datasets (membranes)

  • ~50 grayscale TIFs
  • Each pixel = 6.2 microns
  • Each image is 6.2 microns (one pixel) apart
    • Combine into one 3D matrix

Difficulties Processing Data

  • Background scatter
  • Empty space
  • Fixture touches sample
  • Low contrast!
    • Small range of intensities between:
      • membrane
      • background
      • fixture
  • Difficulty detecting edges

Before and after linear contrasting

img

Intensity Range Histogram

img

Thresholding

  • Set limits to upper and lower values
  • Helps smooth data in background
    • Less edges detected

img

img

Edge Detection

  • Found edges in all three directions
  • Summed gradient values
  • Create separate binary edge matrix

img

img

Noise Elimination

  • Sum of edge in spherical kernel around voxel
  • Threshold based upon surface value (R = 1, <5)
  • Update binary edge matrix
  • Increase then decrease radius

img

img

Closing around Solids

  • Large amount of edges found in solids
  • Closing image to highlight solids
  • Working on statistical means to determine proper radius

img

img

Masking Image

  • Use binary matrix to mask data
  • Highlight the solids (membrane and fixture)

img

Next Steps

Issues with some of the data files…

  • Orientation
  • Cropping
  • Containments

img

Other issues to address

  • Filter out Fixtures
  • Phase identification
  • Corrections for curvature

img

Thank you!

Please look for the “Presentation as Post for Comments” post to add comments or ask questions!

Print

Back