Re: Updated 3D particle counter

Posted by Michael Doube on Dec 30, 2009; 8:12pm
URL: http://imagej.273.s1.nabble.com/3D-shape-descriptors-tp3689894p3689897.html

Hi David

Thanks for reminding me.  I tried and failed (at least, gave up for the
time being to do other things).

Mainly I get tired porting from MATLAB because MATLAB handles matrices
very neatly while Java is somewhat more circuitous, so there's a lot of
tiresome array iterating to do what can be done in 3 characters in an
m-file.

If anyone is interested, the methods I'm in the middle of porting are this:
http://www.mathworks.co.uk/matlabcentral/fileexchange/22683-ellipse-fit-taubin-method
and this:
http://www.mathworks.co.uk/matlabcentral/fileexchange/22684-ellipse-fit-direct-method

Ping me again sometime soon and I'll have another go.

Right, time for tea.

Mike

David Webster wrote:

> Michael,
>
> Have you updated your 2D Circle Fit plugin for ellipses yet?
>
> David Webster
>
> On Wed, Dec 30, 2009 at 11:49 AM, Michael Doube <[hidden email]>wrote:
>
>> Hi all
>>
>> In response to requests here and in the real world (if university counts as
>> the real world), I've made a few adaptations to the 3D Object Counter so
>> that it can do a few fancy new tricks:
>>
>> 1) Label particles a lot faster.  Recursion in particle labelling is
>> reduced and multithreaded so that jobs on big stacks end before our universe
>> does.
>>
>> 2) Euler calculations for each particle, returning Euler characteristic,
>> number of cavities and holes per particle.
>>
>> 3) Surfaces and surface area estimates of each particle
>>
>> 4) Ellipsoid fitting, so that people can work out the
>> oblate/prolate/scalene/spheroidal nature of their particles
>>
>> 5) Moments of inertia and orientation of principal axes
>>
>> 6) Thickness of each particle
>>
>> 7) Display graphical results in a 3D Viewer window, where you can select
>> each numbered particle
>>
>> 8) Display numerical results in a Results table
>>
>> All calculations are done in 3D, and voxel methods are used except for
>> surface area and ellipsoid fitting, which use the results of Bene's surface
>> mesher.
>>
>> At the moment it works on binary stacks, and I've only tested it on the
>> images I have to hand, so your feedback would be very valuable.
>>
>> The plugin is part of BoneJ, which is available here:
>> http://doube.org/bonej.html
>>
>> Once installed, there is a new item in Plugins->Analyze->Particle Analyser.
>>
>> Have a happy new year,
>>
>> Michael