On Tue, Jul 16, 2013 at 9:03 AM, Chintak Sheth <chintaksheth@gmail.com>wrote:

Hi Valeriy,

Wow a compression technique! That would surely be a great addition.

Compression is cool no doubt, but I need it to turn the labeled shape into a polygon =)

...

I am studying a shape descriptors and would like to compute a chain code description of a labeled shape. I was trying to look into docs but have not find a direct way to get this region property with skimage. Is there any way to do it in a few calls which I missed?

- Extract the boundary. `skimage.morphology.erosion` using `skimage.morphology.disk` structuring element with radius 1. Then subtract this from the original image.

I suppose `skimage.morphology.dilation` would work better. It will work in case we computing chain code for the line of 1-pixel width.

- Label the boundaries. If you already have a labeled image, then that saves you a function call else there is `label` function in `skimage.morphology`.

- Use `np.transpose(np.where(label == 1))` to generate the indices of pixels marked as label 1. This will return a `list` of indices. You can simply use the first as the starting point.

This should help set up the stage for the traversal. Hope this helps a bit.

The traverse is the most interesting part in this question =) The other things do not introduce order on the pixels and could be made with current skimage easily.

Chintak.

-- You received this message because you are subscribed to the Google Groups "scikit-image" group. To unsubscribe from this group and stop receiving emails from it, send an email to scikit-image+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.

-- Best regards, Valeriy Sokolov.