Difference between revisions of "2015-01-26 - TrackMate v2.7.1 released"

m
 
Line 6: Line 6:
 
The LAP trackers previously available in TrackMate had a cost calculation based on the square distance between two particles. This kind of costs dealt specifically with Brownian motion. TrackMate allowed to tune these costs with numerical features differences to better deal with motion model that were not brownian or dense particles.  
 
The LAP trackers previously available in TrackMate had a cost calculation based on the square distance between two particles. This kind of costs dealt specifically with Brownian motion. TrackMate allowed to tune these costs with numerical features differences to better deal with motion model that were not brownian or dense particles.  
  
But a tracker that deals specifically with a new motion model is welcome, in particular when it is as common as linear motion. Its principles is detailed  [[TrackMate_algorithms#Linear_motion_tracker|here]].
+
But a tracker that deals specifically with a new motion model is welcome, in particular when it is as common as linear motion. Its principles are detailed  [[TrackMate_algorithms#Linear_motion_tracker|here]].
  
 
[[File:TrackMate KalmanTrackerPrinciple.png|600px]]
 
[[File:TrackMate KalmanTrackerPrinciple.png|600px]]

Latest revision as of 08:44, 27 October 2015

We just release TrackMate v2.7.1 today. The unusual jumps in version number correspond again to internal changes in the core libraries that support the Fiji project.

This version ships a new tracker algorithm for TrackMate, that can deal specifically with particles with a roughly constant velocity, the linear motion LAP tracker.

The LAP trackers previously available in TrackMate had a cost calculation based on the square distance between two particles. This kind of costs dealt specifically with Brownian motion. TrackMate allowed to tune these costs with numerical features differences to better deal with motion model that were not brownian or dense particles.

But a tracker that deals specifically with a new motion model is welcome, in particular when it is as common as linear motion. Its principles are detailed here.

TrackMate KalmanTrackerPrinciple.png



To assess the performance of the new and old trackers, we re-run the ISBI tracking challenge from 2012[1]. We put all the results on a single page: TrackMate Accuracy that should help you decide what is the algorithm that suits your problem best. For instance:



TrackMate RECEPTOR LAP Brownian motion, Linear motion tracker & Nearest neighbor.png



  1. Chenouard et al., "Objective comparison of particle tracking methods", Nature Methods, 2014