Thinning Algorithm:
This program performs skeletonization on the volume passed in the parameter <volfile> (see below)  and creates an mskel file containing the multi-level skeleton. All voxels having values at least as large as the threshold <thresh> (see below) are considered for the skeletonization process. The skeleton produced is in Eucledian metric, and is of multi-resolution format, as is the convention for the mskel format.

Download the source for thinning algorithm here.
Linux executable can be downloaded here.

Usage: The source should compile with standard C++ library. There are no external dependencies.
            euclidskel is the executable name.
 

On the command prompt, do:

    $ ./euclidskel
Usage: euclidskel <volfile> <xs> <ys> <zs> <outfile> <thresh> [measureTimeFlag].
Extract all voxels whose DT-MNT is greater than thresh and create an mskel file.

        <volfile> - volumetric dataset in raw format (array of unsigned char)
        <xs> - size of volume along x-axis.
        <ys> - size of volume along y-axis.
        <zs> - size of volume along z-axis.
        <outfile> - Output file ( explained in detail below)
        <thresh> - All voxels with DT-MNT > thresh are extracted where DT is distance transform of the voxel under   consideration and MNT is mean of its neighboring voxels' distance transforms. This is to ensure that only voxels critical for reconstruction are extracted.
    [measureTimeFlag] - For tracking the time (optional).
 

    The Output file has following format:

    X      Y     Z     DT               DT-MNT or TP

    5       18     0     1.414214       0.952675
    64     18     0     2.000000     1.162645
    74     18     0     2.000000     1.059737
    75     18     0     2.000000     1.031582
    76     18     0     2.000000     0.966883
    84     18     0     2.000000     0.928422
     .......
        .......

                                    X, Y, Z - voxel coordinates
                                    DT = Distance Transform
                                    MNT = mean of DTs of neighoring voxels.
                                    TP = thinness parameter
 

The Skelselect Interface:
 

Skelselect Interface
                       Interface shows visible human skeleton.

Download skelselect interface source for windows here.