I implement a pyopencl kernel to process 3D micrscopy images with CLAHE (Contrast Limited Adaptive Histogram Equalization) with GPU acceleration. I compare its performances with the CPU clahe implementation from skimage (skimage.exposure.equalize_adapthist). I also compare the results to verify that CPU and GPU versions of clahe produce similar equalization.
To test the kernel with your own data, download the notebook and kernel, and add 8-bit 3D stacks named "*raw.tif" to the /Images folder.