Re:Curve Matcher
by wanderfowl on Friday March 15, 2019 @10:27PM
(#19406030)
Out of curiosity, how is this better than, for instance, using B-splines or some other smoothing spline to fit a given chunk of data? I'm not being snarky, I'm just trying to get my head around the use cases for this.
Re:Curve Matcher
by chanind on Saturday March 16, 2019 @06:56AM
(#19407228)
The goal of this library isn't to fit data, but rather to determine if 2 curves have similar shape to each other. So, for instance, if want to determine if a hand-drawn curve has a similar shape to the letter "S" you could use this library to find a similarity score between the curves.
Re:Curve Matcher
by mkl on Friday March 15, 2019 @06:10PM
(#19404785)
This looks neat! I found the Hanzi Writer demo, but you might want to put some (static) examples in the ReadMe, to explain exactly what the library does.
Re:Curve Matcher
by foxes on Sunday March 17, 2019 @09:57PM
(#19418145)
Reminds me of some interesting papers that are related to this topic, off the top of my head check out [0-1]. Perhaps relative geodesics is a key word to search. If you picture curves as made out of some sort of pliable material, you can ask how much energy does it take to deform one curve into another.
Re:Curve Matcher
by chanind on Friday March 15, 2019 @06:06PM
(#19404765)
The original motivation was for doing Chinese character stroke quizzes in Hanzi Writer [https://chanind.github.io/hanzi-writer]. Beyond that I'm not too sure, but hopefully it's useful for somebody :)
[0] https://arxiv.org/pdf/1305.3572.pdf [1] https://royalsocietypublishing.org/doi/abs/10.1098/rspa.2016...
Any idea what you'd use it for?
[0] https://en.wikipedia.org/wiki/Curve_fitting
[1] https://www.investopedia.com/terms/c/cupandhandle.asp