-
Notifications
You must be signed in to change notification settings - Fork 30
Home
Multiple Hypothesis Tracking (MHT) is a centroid tracking algorithm. The algorithm tracks features represented by a position in space as well as some 'texture' data. The basic premise of the algorithm is that multiple hypothetical tracks (called 'hypotheses') can be made using a time series of features, and that an optimal hypothesis can be determined as more information is made available.
This project is based off of code originally produced by Ingemar J. Cox and is available from his Downloads page. The hosting of this project on GitHub has been approved by Dr. Cox, provided that the original copyright is still applied.
There are three Git tags that would point to the three releases made by Dr. Cox.
- v1.0 is the original release
- v1.1 is an update to allow for compilation by gcc 4.1.3
- v1.2 is another update to allow for compilation by gcc 4.3.3
The purpose of this particular project, which will be actively worked on in the master branch, is to adapt this code to be suitable for use in live tracking of storm-cells identified from radar reflectivity images. This will result in a few subtle changes in the behavior of the original algorithm, particularly with respect to allowing for storms to start appearing at any time, and for querying MHT at any time for its best hypothesis.