Top Banner
A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre
21

A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Jan 04, 2016

Download

Documents

Delphia Richard
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

A particle filter to track multiple objects

Carine Hue

Patrick Pérez and Jean-Pierre Le Cadre

Page 2: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Context Two applications:

Signal processing: target tracking

Image analysis: people tracking

observer

target

bearing

Page 3: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Challenges Generic tracking issues

(re)initialization Clutter (false alarms or background) Occlusions

Specific multiple-object issues Complexity Identity confusion when crossing Varying number of entities

Page 4: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Basic ingredients

Hidden variables (position, velocity, shape of the objects)

Data ( bearings, images at each time)

Estimation of posterior

Dynamics:

Likelihood:

Page 5: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Hidden variables Low-dimensional single-object hidden variable:

Bearings-only tracking:

People tracking:

: Fourier descriptors

: 2D translation vector of the center

Multiple-object hidden variable

Page 6: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Single-object dynamics prior: first-order auto-regressive process

Multi-object dynamics: independent single-object dynamics

Dynamics

time t-1

time t ??

velocity vector

position orfirst Fourier coef.

Page 7: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Bearings-only tracking: highly non-linear wrt the state variable

People tracking: obtained with a motion-based segmentation

How to assign the data to the states ?

Data

Page 8: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Data association Notation: association vector: if is issued from object Association assumptions:

one measurement can originate from one object or from the clutter (false alarms)

one object can produce zero or several measurements at one time Exhaustive enumeration ! NP-hard problem

Solution: probabilistic association:

false alarms

Irisa:Irisa:

Page 9: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Particle filtering

Current cloud at time t

Page 10: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Particle filtering

Propagation by sampling from the dynamic prioror from an importance function based on the data

Page 11: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Particle filtering

Weighting according to data likelihood

Page 12: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Particle filtering

Resampling from the weighted particle set

Page 13: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Particle filtering

Prediction again . . .

Page 14: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Example: bearings-only tracking Particle cloud for one object

Page 15: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Multi-object data likelihood

Estimation of the association probabilities vectorwith a Gibbs sampler

take the average value of

t

Page 16: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Example: bearings-only tracking Measurement equation

Differences between the simulated bearings for each object

Page 17: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Example: bearings-only tracking Estimated association probabilities

Estimated trajectories

Page 18: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Example: people tracking Data likelihood

Page 19: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Example: people tracking

Page 20: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

Conclusion and further work Generic multi-object tracker based on a mix of particle

filteringand Gibbs sampling

suitable for various applications

Drawbacks Costly Fixed number of objects For people tracking, the data association could be avoidedwhen there is no ambiguities

Perspectives Varying number of objects Initialization issue Test with real bearings measurements

Page 21: A particle filter to track multiple objects Carine Hue Patrick Pérez and Jean-Pierre Le Cadre.

The End

Questions?