Adapt function rag_mean_color to multi-band rasters (remote sensing research)

[kimzed]

For the rag_mean_color function, I had to modify the function to adapt it to raster data, that can have more than three bands.
I thought with a bit of control flow it would be possible to extend the function to any number of channels.

  for n in graph:
      #graph.nodes[n].update({'labels': [n],
      #                       'pixel count': 0,
      #                       'total color': np.array([0, 0, 0],
      #                                              dtype=np.double)})

      # modification to adapt to multi_label data
      graph.nodes[n].update({'labels': [n],
                             'pixel count': 0,
                             'total color': np.zeros(image.shape[-1],
                                                     dtype=np.double)})

PS: this is my first issue

[rfezzani]

Thank you @kimzed for your interest. This in fact could be a good addition, I also see that rag_mean_color assumes that the channel axis is the last channel. We introduced the channel_axis argument to handle any channel axis position, so it can also be a good update to the function API.

BTW, I will convert this issue to a Discussion as we try to only report bugs in our Issues. We can continue this discussion there ;)