GPU accelerated image processing for everyone
Applies a minimum filter with kernel size 3x3 n times to an image iteratively.
Odd iterations are done with box neighborhood, even iterations with a diamond. Thus, with n > 2, the filter shape is an octagon. The given number of iterations makes the filter result very similar to minimum sphere. Approximately:radius = iterations - 2
Category: Filter
Availability: Available in Fiji by activating the update sites clij and clij2. This function is part of clij2_-2.5.0.1.jar.
Ext.CLIJ2_minimumOctagon(Image input, Image destination, Number iterations);
// init CLIJ and GPU import net.haesleinhuepf.clij2.CLIJ2; import net.haesleinhuepf.clij.clearcl.ClearCLBuffer; CLIJ2 clij2 = CLIJ2.getInstance(); // get input parameters ClearCLBuffer input = clij2.push(inputImagePlus); destination = clij2.create(input); int iterations = 10;
// Execute operation on GPU clij2.minimumOctagon(input, destination, iterations);
// show result destinationImagePlus = clij2.pull(destination); destinationImagePlus.show(); // cleanup memory on GPU clij2.release(input); clij2.release(destination);
% init CLIJ and GPU clij2 = init_clatlab(); % get input parameters input = clij2.pushMat(input_matrix); destination = clij2.create(input); iterations = 10;
% Execute operation on GPU clij2.minimumOctagon(input, destination, iterations);
% show result destination = clij2.pullMat(destination) % cleanup memory on GPU clij2.release(input); clij2.release(destination);
// init CLIJ and GPU importClass(net.haesleinhuepf.clicy.CLICY); importClass(Packages.icy.main.Icy); clij2 = CLICY.getInstance(); // get input parameters input_sequence = getSequence(); input = clij2.pushSequence(input_sequence); destination = clij2.create(input); iterations = 10;
// Execute operation on GPU clij2.minimumOctagon(input, destination, iterations);
// show result destination_sequence = clij2.pullSequence(destination) Icy.addSequence(destination_sequence); // cleanup memory on GPU clij2.release(input); clij2.release(destination);