Hi Tony and Johannes, in fact visualize_boundaries works fine and fulfills all I want, so I will stick with it. The bounding box approach seems fine too! Thanks to both of you for the quick response and the helpful hints! Cheers, Frank On Thursday, November 8, 2012 6:38:56 PM UTC+1, Tony S Yu wrote:

On Thu, Nov 8, 2012 at 4:30 AM, Frank <pennek...@googlemail.com<javascript:>

...

wrote:

...

Hi,

I want to use Python and scikit image for detection, counting and measuring cells from digital pictures. So far most of the things I want to do work fine: I just use a global threshold which separates my cells well from the background. After labelling the thresholded objects the regionprops function provides many of the features I am interested in (area, centroid *etc*). To be completely satisfied, I would like to produce an overlay between the original image and the identified objects after thresholding for error checking. I searched a while to find the proper function to do so and then encountered the mark_boundaries function. However, that one is not working, because the function is not found after importing the skimage.segmentation module (ImportError: cannot import name mark_boundaries). Do you have any suggestions, why it is not working? Or maybe you know a better way to achieve my goal?

Many thanks,

Frank

Hi Frank,

Are you using the latest release (0.7) or the development version on github? It's a bit unfortunate, but the documentation link on the website goes directly to the dev docs instead of the latest release.

I believe `mark_boundaries` is only in the development version of scikit-image, but really that was just a slight modification of `visualize_boundaries`, which should be available in 0.7. Note: `visualize_boundaries` doesn't work with grayscale images, so you may need to call `skimage.color.gray2rgb`.

Hope that helps.

-Tony