search

bmwcarit / barefoot

Java map matching library for integrating the map into software and services with state-of-the-art online and offline map matching that can be used stand-alone and in the cloud.
Apache License 2.0
671 stars 186 forks source link

Incorrect matching of noisy data in v-0.1.1 #69

Closed ghost closed 7 years ago

ghost commented 7 years ago

Hi Sebastian, I have a problem matching some bad quality GPS data with version 0.1.1. Both tests are made with tracker server and version 0.1.0 deals with this perfectly. But in the new version there are same false loop on near streets. Please take a look on the images. I'm attaching the source data also. v-0 1 0 v-0 1 1 points.txt

smattheis commented 7 years ago

I will do a little debug tour to see if there is a bug or if it's actually the noise which needs a little tuning of the parameters. BTW +1 for providing the data to let me debug it.

smattheis commented 7 years ago

In a first test, I haven't found any bug and I got better matching results if I simply increase matcher.sigma to around 10 - 20 meters. The only mismatch is then around the stadium. How does that work for you?

ghost commented 7 years ago

Yes, setting sigma to 15 produces much better results with one break, but it is normal when the GPS positions are far from the road. The important thing is that there are no false paths. Thank you for your help.