ImageJ_javaaddpath

[kouichi-c-nakamura]

This is only a suggestion.

ImageJ has been split into ImageJ and ImageJ_javaaddpath. ImageJ calls ImageJ_javaaddpath to add java class paths.

I have tested that the following code successfully launches ImageJ.

addpath('D:\\Fiji.app\scripts');
ImageJ

In addition, you can use ImageJ_javaaddpath to access ImageJ Java API from within parfor loop of parallel computing MATLAB toolbox.

https://forum.image.sc/t/can-i-use-fiji-imagej-from-within-matlabs-parfor-loop-for-parallel-computing/19007/4?u=kouichi-c-nakamura

fijipath = 'D\Fiji.app\scripts';

addpath(fijipath)
ImageJ

I = imread('corn.tif',3); % uint8

parfor i = 1:10
    addpath(fijipath) % to make MATLAB functions for ImageJ available
    ImageJ_javaaddpath % to make ImageJ Java class available

    I1 = I - uint8(10*i);
    imp = copytoImagePlus(I1);

    disp(imp)

    local_saveasTIFF(imp,i) % you cannot directly access ij.IJ in parfor loop
end

% a local function is needed to access ij.IJ
function local_saveasTIFF(imp,i)
    ij.IJ.saveA
    

In addition, the following code successfully showed an image in ImageJ without ImageJ GUI. I wonder this can be effectively called 'headless' mode. If this disqualifies to be called headless, what else is needed? IJM, perhaps?

addpath('D:\\Fiji.app\scripts');
ImageJ_javaaddpath

A = uint16( randn(100,100)* double(intmax('uint16')));
imp = copytoImagePlus(A)
imp.show()

Above is equivalent to the code below, actually, if this can be called as 'headless'.

addpath('D:\\Fiji.app\scripts');
ImageJ(false)

A = uint16( randn(100,100)* double(intmax('uint16')));
imp = copytoImagePlus(A)
imp.show()

In fact, if yes, we won't need ImageJ_javaaddpath to be a separate function. We can simply work with ImageJ and ImageJ(false).

[imagejan]

Merged in 8d0d50d, closing now.