Stefan van der Walt

Another user report: https://mail.python.org/pipermail/scikit-image/2017-June/005260.html

The generally accepted angle convention is (I think) angle measured anti-clockwise from positive X-axis. @jni Are you still planning on doing a `.transform_xy` module? My feeling is ellipse should be...

Are you suggesting that we always ensure that `a` is larger than `b`, and then modify the angle appropriately, for consistency? There are no guarantees on the a/b order currently,...

> Is there a way to estimate the errors in a and b using this model? You can devise some overall error metric by measuring the overlap in the surface...

This is the general idea: https://stackoverflow.com/a/28287741/214686

One confusing thing about this example: why can you provide the same function (otsu) and get different high/low threshold values out?

@jmetz That's a neat method, but all this information should be in the docs.

It's hard to make this general, because all existing methods imply a certain order of thresholding (high, then low). What if some method comes along that does it the opposite...

@jmetz Do you think the robust thresholding method can be implemented (perhaps not with ideal efficiently, but still) by calling the existing threshold method multiple times? If so, the complexity...

Hi @kolibril13 :wave: Thanks for your interest in scikit-image, and for providing this proof-of-concept front-page widget to demonstrate functionality. The interactive approach is fun and friendly, but given that this...