Presentation Gait Kjetil-holien

Embed Size (px)

Citation preview

  • 8/4/2019 Presentation Gait Kjetil-holien

    1/29

    Gait recognition under non-

    standard circumstances

    Kjetil Holien

  • 8/4/2019 Presentation Gait Kjetil-holien

    2/29

    Disposition

    Research questions

    Introduction

    Gait as a biometric feature

    Analysis

    Experiment setup

    Results

    Conclusion

    Questions

    1/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    3/29

    Research questions

    Main research questions: To what extent is it possible to recognize a person

    under different circumstances?

    Do the different circumstances have any commonfeatures?

    Sub research question: Do people walk in the same way given the same

    circumstances?

    2/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    4/29

    Introduction

    Authentication can occur in three ways: Something you know, password or PIN code.

    Something you has, key or smartcard.

    Something you are, biometrics.

    Biometrics are divided into: Physiological: properties that normally do not change,

    fingerprints and iris. Behavioral: properties that are learned, such as

    signature and gait.

    3/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    5/29

    Gait as a biometric feature

    Three main approaches: Machine Vision based.

    Floor Sensor based.

    Wearable Sensor based (our approach).

    4/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    6/29

    Machine Vision

    Obtained from the distance

    Image/video processing

    Unobtrusive

    Surveillance and

    forensic applications

    5/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    7/29

    Floor Sensor

    Sensors on the floor

    Ground reaction forces/

    heel-to-toe ratio

    Unobtrusive

    Identification

    6/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    8/29

    Wearable sensors

    Sensor attached to the body

    Measure acceleration

    Signal processing

    Unobtrusive

    Authentication

    7/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    9/29

    Performances of related work

    Body location EER, % Number ofSubjects

    Ankle ~ 5 21

    Arm ~ 10 30

    Hip (our approach) ~ 13 100

    Trousers pocket ~ 7.3 50

    8/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    10/29

    Gait analysis

    Sensor records acceleration in three directions: X (horizontal)

    Y (vertical)

    Z (lateral) Average cycle method:

    Detect cycles within a walk.

    A cycle consist of a doublestep (left+right).

    Average the detected cycles (e.g. mean, median). Compute distance between average cycles.

    Euclidian, Manhattan, DTW, derivatitve

    9/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    11/29

    Average cycle method

    Compute resultant vector:

    Time interpolation: every 1/100th sec

    Noise reduction: Weighted Moving Average Step detection

    Average cycle creation

    10/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    12/29

    Raw data, resultant vector

    11/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    13/29

    Time interpolation and noisereduction

    12/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    14/29

    Step detection (1/2)

    13/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    15/29

    Step detection (2/2)

    Consist of several sub-phases: Estimate cycle length

    Indicate amplitude details

    Detect starting location Detect rest of the steps

    14/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    16/29

    Creation of average cycle

    Pre-processing methods: Normalize to 100 samples

    Adjust acceleration

    Align maximum points Normalize amplitude

    Skip irregular cycles

    Create average cycle:

    Mean Median

    Trimmed Mean

    Dynamic Time Warping

    15/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    17/29

    Cycles overlaid

    16/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    18/29

    Average cycle, mean

    17/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    19/29

    Experiment setup

    Main experiment: 60 participants, two sessions of collection.

    1st session: 6 normal walks, 8 fast and 8 slow.

    2nd session: 6 normal walks, 8 circle walks (4 left and 4 right).

    Sub-experiment: 5 participants walking 40 sessions 2 months.

    Each session consisted of 4 walks in the morning and 4 walks inthe evening.

    Sensor was always at the left hip.

    18/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    20/29

    Results

    Best results when: Normalize to 100 samples.

    Adjust acceleration.

    Aligned maximum points. Removed irregular cycles.

    Mean and median average cycle.

    Dynamic Time Warping as distance metric.

    19/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    21/29

    Normal walking

    EER, %

    Automatically Manually1st session 1.64 0.66

    2nd session 1.94 1.04

    All normal 5.91 4.02

    20/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    22/29

    Other circumstances

    EER, %

    Automatically Manually

    Circle left 2.97 1.31Circle right 5.96 0.90

    Fast 3.23 2.94

    Slow 10.71 4.80

    21/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    23/29

    All circumstances

    Normal vs other circumstances EER between 15-30%

    Multi-template 1 template for each circumstance, the others as input

    EER = 5.05%

    22/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    24/29

    Common features

    Cycle length: Normal: [95..125], average of 109 samples

    Fast: [80..110], average of 96 samples

    Slow: [110..180], average of 137 samples Circle same as normal

    Amplitudes related to cycle length

    23/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    25/29

    Long-term experiment (1/3)

    Morning vs morning / evening vs evening Compare sessions at different days intervals

    24/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    26/29

    Long-term experiment (2/3)

    Linear regression to compute a linearfunction (y = a + bx).

    Use hypothesis testing:

    H0: b = 0 (stable walk)

    H1: b > 0 (more unstable walk)

    Results: Rejected H0 for 90% distance increases as time

    passes by.

    25/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    27/29

    Long-term experiment (3/3)

    Morning vs evening (same day) andevening vs the consecutive morning No difference in the average scores.

    Between 30% and 70% increase compared with 1day interval scores.

    26/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    28/29

    Conclusion

    Extremely good EER when comparing thecircumstance with itself.

    Different circumstances seems to bedistinct hard to transform X to normal.

    Good results when using a multi-templatesolution.

    Gait seems to be unstable to some extent need a dynamic template.

    27/27

  • 8/4/2019 Presentation Gait Kjetil-holien

    29/29

    Questions?

    Thanks for listening!