talk 2012-icmew-perception
DESCRIPTION
TRANSCRIPT
![Page 1: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/1.jpg)
Information Technology
Robust Background Subtraction Based on
Perceptual Mixture-of-Gaussians with
Dynamic Adaptation Speed
Mahfuzul Haque and Manzur Murshed
![Page 2: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/2.jpg)
Agenda
Background Subtraction
Statistical Background Subtraction
Perception Inspired Background Subtraction
Dynamic Adaptation Speed
Experiments
Summary
Q&A
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 2
![Page 3: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/3.jpg)
Background Subtraction
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 3
Input
Output
![Page 4: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/4.jpg)
Background Subtraction: Challenges
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 4
Illumination variation
Local background motion
Camera displacement
Shadow and reflection
Challenges
Current frame
Background
Model
Foreground Blob
Dynamic Background Subtraction(e.g., MOG)
Basic Background Subtraction (e.g., BBS)
- =
Current frame Background Foreground Blob
![Page 5: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/5.jpg)
Statistical Background Subtraction
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 5
Te
P(x)
x x
MOG: x = c1σ
μ
P(x)
x x
b
x = c2b
ω1
σ12
µ1
road
ω2
σ22
µ2
shadow
ω3
σ32
µ3
car
65% 20% 15%
BBS: x = c
Statistical Approaches Our Hypothesis (Perception Inspired)
![Page 6: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/6.jpg)
Perception Inspired Background Subtraction
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 6
P(x)
x x
b
x = c2b Detection with
Low x
Current
Frame
Detection with
High x
![Page 7: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/7.jpg)
Weber’s Law
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 7
Ernst Weber, an experimental psychologist in the 19th
century, observed that the just-noticeable increment ΔI
is linearly proportional to the background intensity I.
ΔI = c2I
How human visual system perceives noticeable intensity
deviation from the background?
![Page 8: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/8.jpg)
Weber’s Law
Te
Ernst Weber, an experimental psychologist in the
19th century, observed that the just-noticeable
increment ΔI is linearly proportional to the
background intensity I.
P(x)
x x
b ? x b
ΔI = c2I
December 30, 2013 8 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed
x = c2b
![Page 9: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/9.jpg)
Perceptual tolerance of HVS
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 9
Method 1
Method 2 Reference
Image
Distorted
Images
p dB
q dB |p – q| < 0.5 dB
Not perceivable
by human visual
system
What is the perceptual tolerance level in distinguishing
distorted intensity measures?
![Page 10: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/10.jpg)
Our Problem: c2 = ?
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 10
Te
x = c2b
P(x)
x x
b
Weber’s Law
Perceptual Threshold, TP (0.5 dB)
1255
10log20255
10log20
xbxb2TP
x = c2b
![Page 11: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/11.jpg)
Linear Relationship
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 11
x
b
![Page 12: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/12.jpg)
Rod and Cone
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 12
Rods and Cones are two different types of
photoreceptor cells in the retina of human eye
Rods
– Operate in less intense light
– Responsible for scotopic vision (night vision)
Cones
– Operate in relatively bright light
– Responsible for photopic (color vision)
![Page 13: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/13.jpg)
Error Sensitivity in Darker Background
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 13
![Page 14: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/14.jpg)
Piece-wise Liner Relationship
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 14
Te
Scotopic Vision (R) Photopic Vision (C)
![Page 15: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/15.jpg)
Dynamic Adaptation Speed
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 15
•Sleeping person problem
•Walking person problem
![Page 16: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/16.jpg)
Dynamic Adaptation Speed
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 16
![Page 17: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/17.jpg)
Dynamic Adaptation Speed
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 17
![Page 18: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/18.jpg)
Dynamic Adaptation Speed
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 18
![Page 19: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/19.jpg)
Experiments
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 19
Total 50 test sequences from 8 different sources
Scenario distribution
Indoor Outdoor Multimodal Shadow and Reflection Low background-foreground contrast
Test Sequences
Evaluation
Qualitative and quantitative comparison:
MOG (S&G) (TPAMI, 2000)
MOG (Lee) (TPAMI, 2005)
ViBe (TIP, 2011)
False Positive (FP)
False Negative (FN)
False Classification
![Page 20: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/20.jpg)
Test Sequences
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 20
Te PETS (9) Wallflower (7) UCF (7) IBM (11) CAVIAR (7) VSSN06 (7) Other (2)
![Page 21: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/21.jpg)
Experiments
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 21
![Page 22: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/22.jpg)
Experiments
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 22
![Page 23: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/23.jpg)
Experiments
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 23
![Page 24: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/24.jpg)
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 24
First
Frame
Test
Frame
Ground
Truth
MOG
(S&G)
MOG
(Lee)
ViBe Proposed
![Page 25: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/25.jpg)
Summary
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 25
Realistic background value prediction: high model agility
and superior detection quality at fast learning rate.
No context related information: high stability across
changing scenarios.
Perception based detection threshold: superior detection
quality in terms of shadow, noise, and reflection.
Perceptual model similarity: optimal number of models
throughout the system life cycle.
Parameter-less background subtraction: ideal for real-
time video analytics.
![Page 26: Talk 2012-icmew-perception](https://reader033.vdocument.in/reader033/viewer/2022051514/54842ea0b47959140d8b4b38/html5/thumbnails/26.jpg)
Q&A
December 30, 2013 Perceptual Mixture-of-Gaussians with Dynamic Adaptation Speed 26