Research | People | Publications | Software | Data | Vacancies

Space-timE RePresentation, Imaging

and cellular dynamics of

molecular COmplexes




3D Optical flow visualization

The following links are provided as supplementary materials to the main article. First, we provide links to run 3DPatchMatch and variational method online. You have to upload source, target and source mask files. After the programs have run successfully, you can have for:

    • U,V,W scalar fields as 3D tif file which are necessary as inputs to visualization scripts (described later),
    • Magnitude file as 3D tif file.
    • Each of these files is available for each coarse-to-fine levels and can be optionally viewed in your fiji/imageJ software once download.
  • Variational Method (link: https://allgo18.inria.fr/apps/3dctvar)
    • U,V,W scalar fields as 3D tif file which are necessary as inputs to visualization scripts (described later),
    • Magnitude file as 3D tif file.
    • You can input U,V,W files obtained after 3DPatchMatch in addition to source and target files (no mask required for this). On doing so, the output will be combination of 3DPatchMatch and variational refinement. You should do so only when the original sequence has large motion.

Note: Please register at gitlab to use the software online.

Now, we provide a link to download a zipped folder containing matlab scripts for:

  • 3DHSV visualization
  • 3PHS visualization
  • computation of WSAE measure
  • sample real sequence (collagen motion) to compute flow fields

All of these scripts require the output of aforementioned programs. WSAE scripts requires additionally, the source, the target and the source mask files.

link : https://gitlab.inria.fr/serpico/3d-flow-assessment/

Now, we provide the videos of sequences described in the paper. The following videos (available in .avi and .tif format) contain 3D Optical flow visualization of:

For additional information/help contact: sandeep.manandhar@inria.fr


Spot in M10 dataset

This dataset contains 6 collections of 100 images. As explained in [1], 100 backgrounds were extracted from real TIRFM images. Then Gaussian spots where added, as well as Poisson-Gaussian noise.

The collections were obtained by varying spot variance and PSNR as follows:

  • M10-1-18 Spot variance: 1; PNSR: 18.
  • M10-1-21 Spot variance: 1; PNSR: 21.
  • M10-1-23 Spot variance: 1; PNSR: 23.
  • M10-1-25 Spot variance: 1; PNSR: 25.
  • M10-1-30 Spot variance: 1; PNSR: 30.
  • M10-1.44-30 Spot variance: 1.44; PNSR: 30.

Please cite [1] when referring to this dataset.

[1] A. Basset, J. Boulanger, J. Salamero, P. Bouthemy, C. Kervrann. Adaptive spot detection with optimal scale selection in fluorescence microscopy images, IEEE Transactions on Image Processing, 2015.