On Mon, 29. Jun 09:26, Valery Kot wrote: > On Sun, Jun 28, 2020 at 12:03 AM Andriy Gelman <andriy.gel...@gmail.com> > wrote: > > > > lgtm. I saw a small improvement when testing. > > > > Would be nice to allow weak edges that become strong to trigger neighboring > > weak > > edges in the future. > > Thanks for the comment. > > You are right, ideally double_threshold should be recursive (seed from > each "high" peak and spread over "low" peaks). But then potentially we > may end up with width*height/2 recursion depth, which may lead to > stack overflow. So probably some recursion limit is needed, and hence > suboptimal solution. > > Or iterative approach, running through the complete image again and > again checking if "low" peaks are touching already selected edge > pixels. Stop when no new pixels can be added to the edge. Better, but > still potentially width*height/2 iterations with width*height > operations each, completely killing performance. > > Maybe later I'll try to implement it in a generic way, but this is out > of scope for this patch. > > Regards, > > Valery Kot
Any objections if I apply this patch? -- Andriy _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
