1 gesture recognition using hmms and size functions

11
1 Gesture recognition Using HMMs and size functions

Upload: cody-freeman

Post on 28-Mar-2015

229 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 1 Gesture recognition Using HMMs and size functions

1

Gesture recognition

Using HMMs and size functions

Page 2: 1 Gesture recognition Using HMMs and size functions

2

Approach

Combination of HMMs (for dynamics) and size functions (for pose representation)

Page 3: 1 Gesture recognition Using HMMs and size functions

3

Size functionsTopological representation of contours

Page 4: 1 Gesture recognition Using HMMs and size functions

4

Measuring functions

Functions on the contour to which the size function is computed

real image

measuring function

family of lines

Page 5: 1 Gesture recognition Using HMMs and size functions

5

Feature extraction 1

An edge map is extracted from the image

real image

edge map

… and …

Page 6: 1 Gesture recognition Using HMMs and size functions

6

Feature extraction 2a family of measuring functions is chosen

… the szfc are computed, and their means form the feature vector

Page 7: 1 Gesture recognition Using HMMs and size functions

7

Hidden Markov modelsFinite-state model of gestures as sequences of a small number of poses

Page 8: 1 Gesture recognition Using HMMs and size functions

8

Four-state HMM

Gesture dynamics -> transition matrix A

Object poses -> state-output matrix C

Page 9: 1 Gesture recognition Using HMMs and size functions

9

EM algorithm

feature matrices: collection of feature vectors along time

EM A,C

learning the model’s parameters through EM

two instances of the same gesture

Page 10: 1 Gesture recognition Using HMMs and size functions

10

EM algorithm -> learning the model’s parameters

Learning algorithm

Page 11: 1 Gesture recognition Using HMMs and size functions

11

Gesture classification

HMM 1

HMM 2

HMM n

the new sequence is fed to the learnt gesture’s models

they produce a likelihoodthe most likely model is chosen (if above a threshold)