effects of post-processing on background subtraction algorithms
DESCRIPTION
Effects of Post-processing on Background Subtraction Algorithms. Donovan Parks. Outline. What is background subtraction? Project motivation How is BGS performed and what makes it difficult? Project goals and results Concluding remarks. What is background subtraction?. - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/1.jpg)
Effects of Post-processing on Background Subtraction Algorithms
Donovan Parks
![Page 2: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/2.jpg)
Outline
What is background subtraction?
Project motivation
How is BGS performed and what makes it difficult?
Project goals and results
Concluding remarks
![Page 3: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/3.jpg)
What is background subtraction?
Real-time method for identifying moving foreground objects within a video
![Page 4: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/4.jpg)
Project motivation BGS is an important low-level step in
many computer vision applications: Video surveillance Traffic monitoring FG/BG segmentation
My interest is in using BGS to extract human silhouettes for pose estimation
How “good” are the obtained silhouettes in unconstrained environment?
Images from: Sminchisescu and Telea, “Human Pose Estimation from Silhouettes”, 2002.
![Page 5: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/5.jpg)
How is BGS performed?
Static frame differencing BG model = first frame of video
BGDifferencing
Input Stream
BG Model
Output Masks
Threshold
![Page 6: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/6.jpg)
What makes BGS difficult?
Moving background elements:
![Page 7: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/7.jpg)
Adaptive, statistical BG models
BGDifferencing
Mean
Input Stream
BG Model
Output Masks
Threshold
Update BGModel
0 50 100 150 200 2500
0.005
0.01
0.015
0.02
0.025
0.03
Component Value
Pro
babi
lity
0 50 100 150 200 2500
0.005
0.01
0.015
0.02
0.025
0.03
Component Value
Pro
babi
lity
0 50 100 150 200 2500
0.005
0.01
0.015
0.02
0.025
0.03
Component Value
Pro
babi
lity
Variance
Gaussian Pixel Model
![Page 8: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/8.jpg)
What makes BGS difficult?
Shadows:
![Page 9: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/9.jpg)
Shadow removal
Shadows have little effect on chromaticity, but reduce luminosity
BGDifferencing
Mean
Input Stream
BG Model
Output Masks
Threshold
Update BGModel
Variance
ShadowRemoval
![Page 10: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/10.jpg)
What makes BGS difficult?
Ghosting:
![Page 11: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/11.jpg)
Ghost detection via optical flow
Low optical flow = ghost!
BGDifferencing
Mean
Input Stream
BG Model
Output Masks
Threshold
Update BGModel
Variance
ShadowRemoval
ConnectedComponents
OpticalFlow Test
![Page 12: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/12.jpg)
What else makes BGS difficult?
FG/BG blending
![Page 13: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/13.jpg)
Project goals
Evaluate a selection of state-of-the-art background subtraction algorithms Considering 10 algorithms in all
Analyze how post-processing influences the performance of these algorithms Shadow removal Optical flow testing Morphological “cleaning” Area thresholding
![Page 14: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/14.jpg)
Initial resultsP
reci
sion
Recall
![Page 15: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/15.jpg)
Example of shadow removal
![Page 16: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/16.jpg)
Example of “cleaned” results
![Page 17: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/17.jpg)
Conclusions
Many factors which make BGS difficult
Post-processing can significantly improve results
Results not as “clean” as more computationally expensive approaches
![Page 18: Effects of Post-processing on Background Subtraction Algorithms](https://reader035.vdocument.in/reader035/viewer/2022062309/5681351a550346895d9c71b3/html5/thumbnails/18.jpg)
Questions?
Thank you.