An image processing example

from fairdatanow import DataViewer, filters, login
import os 
import re 
url = "https://laboppad.nl/falnama-project"
dv = DataViewer(login(url), **filters(search='edited.tif', extensions=['.tif']))
Connection with nextcloud server is ok.

Let’s take a look at the filenames.

im_files = dv.download_filtered()
titles = [re.sub('_Falnama.*', '', os.path.basename(f)) for f in im_files]
Ready with downloading 28 selected remote files to local cache: /home/frank/.cache/fairdatanow                                                                      

The holoviews way

Given these rather big files, we can use the awesome Python package holoviews to create an interactive visualization in a few lines of code in 15 seconds on my machine…

import imageio.v3 as iio
import holoviews as hv
from holoviews.operation.datashader import rasterize 
import panel as pn
from pyvips import Image
hv.extension('bokeh')
ⓘ
pn.extension(sizing_mode='stretch_width')
# Read the 28 super large tif files into arrays with pyvips 
ims = [Image.new_from_file(im_file).numpy() for im_file in im_files] 

# Convert into holoviews images
hv_images = [hv.Image(im) for im in ims] 


# Rasterize to reduce size and speed up loading 
rasterized_ims = [rasterize(hv_im, width=500, height=500) for hv_im in hv_images] 


# Embed images in interactive viewer 
image_panes = [pn.pane.HoloViews(raster_im.opts(xaxis=None, yaxis=None, cmap='gray', tools=['hover'], title=titles[i])) for i, raster_im in enumerate(rasterized_ims)]
# Create a grid 
#xrays = pn.panel(pn.GridBox(*image_panes, ncols=6))

xrays = pn.GridBox(*image_panes, ncols=6)

xrays

This is quite nice. We might improve this code by soft coding the height and width of the thumbnails somehow.