Upload
holly
View
52
Download
0
Tags:
Embed Size (px)
DESCRIPTION
Images for paper. By shooting. Sample collection. Hard/Soft vertical touch. timer. 5. Finger touch position. Background Subtraction. Normalization. Difference of Gaussian. Binary. Finger Analysis. Background Subtraction. Normalization. Approximate DoG. Binary. Finger Analysis. - PowerPoint PPT Presentation
Citation preview
Images for paper
By shooting
2
Sample collection
• Hard/Soft vertical touch
Finger touch position
5
timer
3
BackgroundSubtraction Normalization Simple
HighpassMono Threshold
BackgroundSubtraction Normalization Difference of
Gaussian Threshold
Normalization Difference of Gaussian
Background Subtraction Binary Finger Analysis
Normalization Difference of Gaussian
Background Subtraction Binary Finger Analysis
Normalization ApproximateDoG
Background Subtraction Binary Finger Analysis
4
ImageFusion(Blend)
IR Camera
IR camera
Undistortion
Undistortion
HomoWarp
HomoWarp
ImageFusion
(Stitching)
IR Camera
IR camera
Undistortion
Undistortion
HomoWarp
HomoWarp
5
DetectionModule Verify Next Parameter Set
Generator
DetectionResult
GroundTruth
Error RateParameter Set
Parameter Set
Parameter Set’
6
7
Detection systemSampleset
Training
Parameter Set
Detection Result
GroundTruth
OptimalParameter
Set
8
VerifyNext Parameter Set
Generator
DetectionResult
GroundTruth
Error Rate
9
Detection systemSampleset
Training
Parameter Set
Detection Result
GroundTruth
OptimalParameter Set
VerifyNext Parameter Set
Generator
DetectionResult
GroundTruth
Error Rate
Parameter Set
10
11
12
Detection systemTest set
Optimal parameter finder
Parameter Combination
Detection Result
Optimal Parameter Combination
Ground truth(Trace)
VerifyNext Parameter Set
Generator
DetectionResult
GroundTruth
# of miss & # of false alarm
Parameter Combination
Optimal parameter finder
Applicable Parameters Determinator
Parameter Combination
Detection Result
Applicable set of Parameters
TestSet
Touch Data
Ground truth(Trace)
Detection System
14
frame 1
frame 2
miss frame 1
frame 2
False alarm
miss
15
frame 1
frame 3
frame 2 frame 1
frame 3
frame 2
frame 3False alarm
miss
16
frame 1
frame 1
False alarm
17
Undistortion
18
Dead zone
19
20
ImageStiching
FingerDetection
Finger Tracking
21
Image blend
23
Hardware configuration
• Order of diffuser layer and touch-glass layer
Diffuser layer
IR illuminatorIR camera
spot
IR illuminatorIR camera
Touch-glass layer
IR camera
spot
IR camera
24
Modified by 510
• Order of diffuser layer and touch-glass layer
Diffuser layer
IR illuminatorIR camera
spot
IR illuminatorIR camera
Touch-glass layer
IR camera
spot
IR camera
2
1
4 3
Sampling
Measure
1 1( | )t tP X Z
1( | )t tP X Z
( | )t tP X Z
INITIALIZE PARTICLES
PARTICLE GENERATION
WEIGHT COMPUTATION
1 2 M
1 2 M
NORMALIZE WEIGHTS
More iteration
Over l imited iteration
RESAMPLING
OUTPUTNo
Yes
Yes
No
……
……
PARTICLE GENERATION
WEIGHT COMPUTATION
5 2
16
2 5
43
1
4
3
6
VSLH VS
RH
2
5
5 2
16
2 5
43
1
4
3
6
VSLH VS
RH
2
5
INITIALIZE PARTICLES
Particle Generation
Weight Computation
1 2 M
1 2 M
Normalize Weights
Mo re I te ra t ion
Over
Resampling
OUTPUTNo
Yes
Yes
No
VSLH
(2) IR Camera
(3) IR Illuminator
(1) Peripheral Projector
Optimal Parameter Finder
Finger Detection System
Interface
GPU
Direct3D
HLSL
Dshow
GShow DirectShow
HomoWarp ImageFusion
BackgroundSubtraction Normalization Difference
of Gaussian Binary FingerAnalysis
Multi-touch Finger Detection System
GShow DirectShow
HomoWarp ImageBlending
BackgroundSubtraction Normalization Difference
of Gaussian Binary FingerAnalysis
Multi-touch Finger Detection System
Normalization Difference of Gaussian
Background Subtraction Binary Finger Analysis
Subtract value
Smooth kernel Threshold Finger size
Low bound 0 7 10 15
Step 3 2 3 9
High bound 15 17 25 60
RGBTo Gray
Background Subtraction Normalization Approximate
DoG Threshold
Frame per second 410 385 400 400