Download - Background Removal
![Page 1: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/1.jpg)
Background Removal
Leow Wee Kheng
CS4243 Computer Vision and Pattern Recognition
CS4243 Background Removal1
![Page 2: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/2.jpg)
Here’s an image…
We often just want the eagle
CS4243 Background Removal 2
Background Removal
![Page 3: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/3.jpg)
Background Removal
Related to tracking and segmentation Tracking
Tracks location of moving object in video. Segmentation
Separate object and background in single image. Background removal
Separate object and background given > 1 image.
CS4243 Background Removal 3
![Page 4: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/4.jpg)
Background Removal
Two general approaches: With known background, also called clean plate. Without known background.
CS4243 Background Removal 4
![Page 5: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/5.jpg)
With Clean Plate
Clean plate: background only image
CS4243 Background Removal 5
![Page 6: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/6.jpg)
Subtract clean plate P from image I
Colour image has 3 components R: red, G: green, B: blue So, get 3 sets of differences
CS4243 Background Removal 6
),(),(),( yxPyxIyxD
),(),(),(
),(),(),(
),(),(),(
yxPyxIyxD
yxPyxIyxD
yxPyxIyxD
BBB
GGG
RRR
absolute difference
![Page 7: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/7.jpg)
Combine 3 sets of differences into 1 set
R, G, B are constant weights.
Usually, R G B 1.
In the case of equal weights, R G B 1/3.
CS4243 Background Removal 7
),(α),(α),(α),( yxDyxDyxDyxD BBGGRR
![Page 8: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/8.jpg)
CS4243 Background Removal 8
imageclean plateabsolute colour differenceabsolute difference
![Page 9: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/9.jpg)
Finally, fill in foreground object colour
is threshold. If D(x, y) > , pixel at (x, y) is foreground pixel. B is constant background colour, e.g., black.
CS4243 Background Removal 9
otherwise
),( if),(),(
B
yxDyxIyxF
![Page 10: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/10.jpg)
CS4243 Background Removal 10
imageclean plateabsolute colour differenceabsolute difference
![Page 11: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/11.jpg)
Notice
Some parts of the eagle’s tail are missing.Why?
CS4243 Background Removal 11
![Page 12: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/12.jpg)
Dynamic Clean Plate
Stationary camera Stationary background. Need only one image as clean plate.
Moving camera Moving background. Need a video clean plate. With motion-controlled camera, controlled lighting
Shoot clean plate video. Shoot target video with same camera motion.
Remove background with corresponding clean plate.
CS4243 Background Removal 12
![Page 13: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/13.jpg)
clean plate
CS4243 Background Removal 13
![Page 14: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/14.jpg)
scene video
CS4243 Background Removal 14
![Page 15: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/15.jpg)
background removed
CS4243 Background Removal 15
![Page 16: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/16.jpg)
Without Clean Plate
Background removal without clean plate is more difficult.
Possible if moving objects do not occupy the same position all the time.
3 cases Stationary camera, fixed lighting. Stationary camera, varying lighting. Moving camera.
CS4243 Background Removal 16
![Page 17: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/17.jpg)
Stationary Camera, Fixed Lighting
Consider these video frames:
Moving object occupies a small area. Moving object does not occupy the same position. What if we average the video frames?
CS4243 Background Removal 17
![Page 18: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/18.jpg)
Averaging
Mean of video frame
i : frame number n : number of frames
Notes: The above direct formula can lead to overflow error. Refer to colour.pdf for a better formula.
CS4243 Background Removal 18
),(1
),( yxIn
yxMi
i
![Page 19: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/19.jpg)
Averaging gives mostly background colours. Some faint foreground colours remain.
CS4243 Background Removal 19
Case 1: average over whole video
![Page 20: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/20.jpg)
Foreground colours are more localised in one region. Foreground colours are stronger.
CS4243 Background Removal 20
Case 2: average over first 3 seconds
![Page 21: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/21.jpg)
Subtract background from video frame
CS4243 Background Removal 21
Case 1 Case 2
![Page 22: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/22.jpg)
Copy foreground colours to foreground pixels
Background colours are removed: true rejection. Some foreground colours are missing: false rejection.
CS4243 Background Removal 22
Case 1 Case 2
![Page 23: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/23.jpg)
Use lower thresholds
More foreground colours are found: true acceptance. Background colours are also found: false acceptance.
CS4243 Background Removal 23
Case 1 Case 2
![Page 24: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/24.jpg)
Another example
CS4243 Background Removal 24
![Page 25: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/25.jpg)
Averaging video frames
CS4243 Background Removal 25
Case 1: over whole video Case 2: over first 3 seconds
![Page 26: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/26.jpg)
Subtract background from video frame
CS4243 Background Removal 26
Case 1 Case 2
![Page 27: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/27.jpg)
Copy foreground colours to foreground pixels
Background colours are removed: true rejection. Some foreground colours are missing: false rejection.
CS4243 Background Removal 27
Case 1 Case 2
![Page 28: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/28.jpg)
Use lower thresholds
More foreground colours are found: true acceptance. Background colours are also found: false acceptance.
CS4243 Background Removal 28
Case 1 Case 2
![Page 29: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/29.jpg)
Background Modelling
Averaging is simple and fast but not perfect.
Better than average: colour distribution. For each pixel location,
compute distribution of colours over whole video.
CS4243 Background Removal 29
![Page 30: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/30.jpg)
For a background pixel:
Single cluster of colours (due to random variation). Peak: most frequent colour.
CS4243 Background Removal 30
![Page 31: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/31.jpg)
For a pixel that is background most of the time:
Two clusters: background, foreground. Relative height: duration covered by foreground.
CS4243 Background Removal 31
![Page 32: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/32.jpg)
k-means clustering
A method for grouping data points into clusters.
Represent each cluster Ci by a cluster centre wi.
Repeatedly distribute data points and update cluster centres.
CS4243 Background Removal 32
![Page 33: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/33.jpg)
k-means clustering
1. Choose k initial cluster centres w1(0),…, wk(0).
2. Repeat until convergence Distribute each colour x to the nearest cluster Ci (t)
Update cluster centres:Compute mean of colours in cluster
CS4243 Background Removal 33
t is iteration number
![Page 34: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/34.jpg)
For background removal, can choose k = 2 One for foreground, one for background.
Initial cluster centres Get from foreground and background in video.
Possible termination criteria Very few colours change clusters. Fixed number of iterations.
After running clustering If foreground area is small, then smaller cluster is
foreground.
CS4243 Background Removal 34
![Page 35: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/35.jpg)
Background removed
Most background colours are removed. A bit of shadow remains. Most foreground colours are found.
CS4243 Background Removal 35
![Page 36: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/36.jpg)
Stationary Camera, Varying Lighting
Basic ideas Multiple background clusters for different lighting
conditions. Apply k-means clustering with k > 2.
CS4243 Background Removal 36
![Page 37: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/37.jpg)
Example from [Stauffer98]
CS4243 Background Removal 37
![Page 38: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/38.jpg)
Moving Camera
Basic ideas Track and recover camera motion [Bergen92]. Stabilise video by removing camera motion
[Matsushita05]. Do stationary camera background removal. Put back camera motion.
CS4243 Background Removal 38
![Page 39: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/39.jpg)
Summary
With clean plate Subtract clean plate from video frames.
Without clean plate Estimate background
Average video frame Cluster pixel colours
Subtract estimated background from video frames.
Moving camera Stabilise video, then perform background removal.
CS4243 Background Removal 39
![Page 40: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/40.jpg)
Further Reading
Code book method OpenCV [Bradski08] chapter 9.
Varying lighting condition [Stauffer98]
Motion estimation [Bergen92]
Video stabilization [Matsushita05]
CS4243 Background Removal 40
![Page 41: Background Removal](https://reader034.vdocuments.us/reader034/viewer/2022051620/56813f17550346895da9ab74/html5/thumbnails/41.jpg)
References G. Bradski and A. Kaebler, Learning OpenCV, O’Reilly, 2008.
J. R. Bergen, P. Anandan, K. J. Hanna, and R. Hingorani. Hierarchical model-based motion estimation. In Proc. ECCV, pages 237–252, 1992.
Y. Matsushita, E. Ofek, X. Tang, and H.Y. Shum. Fullframe video stabilization. In Proc. CVPR, volume 1, pages 50–57, 2005.
C. Stauffer and W. E. L. Grimson. Adaptive background mixture models for real-time tracking. In Proc. IEEE Conf. on CVPR, 1998.
CS4243 Background Removal 41