Gabriel Selzer

Alright, I gave this PR some TLC. @imagejan, after some thinking I realized that I liked your solution a bit better, as I described in 5440061. I found it difficult...

> I think the way setting the seed parameter works should be flipped, i.e. don't use a constant seed unless otherwise specified. Thanks for catching that @rimadoma! Should be updated...

Hey @tinevez! I am really sorry I did not see your messages; my notifications were going to the one email that I do not receive notifications for (I changed it...

@tinevez wrote: > Normally you would threshold the edge image (defined over RealType) to generate a binary image from it, but this has to be a separate operation, lest we...

I had the opportunity to test your implementation against possibly the meanest image that I could think of. I drew 25 circles, some of them more visible than others: ...

Hey @tinevez! So I think with `Contingent` it is better to leave it in and to remove the runtime check. I wrote a pretty detailed post above after discussing it...

@tinevez Here is the code I used. ``` Context c = new Context(); final ImageJ ij = new ImageJ(c); ij.launch( args ); final Context context = new Context(); final OpService...

Tackling other PRs first, so it might be a while before I get back to this, but when I do: * Compare the output of this filter to that of...

@tinevez I finally got around to working again on this PR and I started by comparing the results of this PR with the results of the [UCB Hough Circle Transform...

@ctrueden and I have cleaned up history and rebased over the latest master. @tinevez the issues with detection quality still exist. I am moving on to other projects for now,...