Open voyage19941119 opened 7 years ago
They are all variants on ICP, but with different error terms. The solution generally relies on some form of Gauss-Newton method for least-squares fitting the current observations to a model. I find it easier to read the papers if you want the details, not the code. I'm unsure whether this counts as an actual issue with the software implementation
I have problems in reading codes in ICP parts ,I am a little bit confused about ICP part in Kintinuous ,ElasticFusion , are they same to ICP in kinetcFuison? solved by approximately linear method? Thanks for answer!