On 22 November 2017 23:08:17 Jérôme Kieffer <goo...@terre-adelie.org> wrote:
On Thu, 23 Nov 2017 12:33:21 +1100
Juan Nunez-Iglesias <jni.s...@gmail.com> wrote:
Sounds great! I’ll admit that I’m not confident in this area either,
but my reading of the documentation suggests that this is the right
approach: the damping controls the square norm of x. By keeping it
small (with large damping), you force the elements to be non-negative.
The approach looks OK, I am just a bit "unconfident" about tweeking a
factor to get the image I am expecting. Not very scientific when the
resulting image has to used as input for quantitative analysis.
There are various ways to do this, but you have to dampen/penalize the
solution: oscillatory results often achieve the same norm as smooth
results, and the optimization algorithm otherwise has no way to choose
which is best.
Instead of applying explicit regularization, you can also terminate
conjugate gradients early, for example.
Best regards
Stéfan
_______________________________________________
scikit-image mailing list
scikit-image@python.org
https://mail.python.org/mailman/listinfo/scikit-image
