Background Subtraction

A collection of background subtraction algorithms, implemented on top of Boost and OpenCV. See the COPYING for license rights and limitations (BSD).

Algorithms

Adaptive Median

It implements 1995McFarlane.

Grimson Mixture of Gaussians

Presented in 1999Grimson. The implementation is precisely following the description in the paper.

Zivkovic Mixture of Gaussians

As presented in 2004Zivkovic. It slightly differs from Zivkovic implementation in using just one threshold for testing matching to a distribution.

Type-2 Fuzzy Mixture of Gaussians

As presented in 2008ElBaf-2. The difference from other mixtures of Gaussians is in the use of a fuzzy-logic function to estimate the membership to a Gaussian distribution.

Fuzzy Choquet

It implements 2008ElBaf. It applies the Choquet integral for each pixel of the image, to a triple of floating point values that represent:

similarity between foreground and background LBP at that point, and
similarity between two color planes of foreground and background, respectively.

where the color planes are notably not RGB (the paper uses YCrCb).

Fuzzy Sugeno

It implements 2006Zhang. Like above, but it uses a Sugeno integral in Ohta colorspace.

Sigma-Delta

It implements 2007Manzanera.

Temporal Median

It implements an approximation of 2006Calderara. The divergence is in the use of a simpler pair of masks, with global thresholds.

Utilities

There are a bunch of one-line internal utilities in the library that are useful to me, mostly related to conversion between formats, scaling, etc. Find them in src/utils.cpp.

I copied Eric Niebler's getline range code and adapted it to fetching frames from a cv::VideoCapture object. It allows for some sweetness:

cv::VideoCapture cap = ...
for (const auto& frame : bs::getframes_from (cap)) {
    // ... use the frame
}

The code for that is in include/bs/frame_range.hpp, enjoy.

CMake and Windows

No.

References

[1995McFarlane] McFarlane, Nigel JB, and C. Paddy Schofield. "Segmentation and tracking of piglets in images." Machine vision and applications 8.3 (1995): 187-193.

[2006Zhang] Zhang, Hongxun, and De Xu. "Fusing color and texture features for background model." Fuzzy Systems and Knowledge Discovery (2006): 887-893.

[2007Manzanera] Manzanera, Antoine, and Julien C. Richefeu. "A new motion detection algorithm based on Σ–Δ background estimation." Pattern Recognition Letters 28.3 (2007): 320-328.

[2006Calderara] Calderara, Simone, et al. "Reliable background suppression for complex scenes." Proceedings of the 4th ACM international workshop on Video surveillance and sensor networks. ACM, 2006.

[2008ElBaf] El Baf, Fida, Thierry Bouwmans, and Bertrand Vachon. "Fuzzy integral for moving object detection." Fuzzy Systems, 2008. FUZZ-IEEE 2008.(IEEE World Congress on Computational Intelligence). IEEE International Conference on. IEEE, 2008.

[1999Grimson] Stauffer, Chris, and W. Eric L. Grimson. "Adaptive background mixture models for real-time tracking." Computer Vision and Pattern Recognition, 1999. IEEE Computer Society Conference on. Vol. 2. IEEE, 1999.

[2004Zivkovic] Zivkovic, Zoran. "Improved adaptive Gaussian mixture model for background subtraction." Pattern Recognition, 2004. ICPR 2004. Proceedings of the 17th International Conference on. Vol. 2. IEEE, 2004.

[2008ElBaf-2] Fida El Baf, Thierry Bouwmans, Bertrand Vachon. Type-2 Fuzzy Mixture of Gaussians Model: Application to Background Modeling. G. Bebis et al. (Eds). ISVC 2008, Dec 2008, Las Vegas, United States. Springer-Verlag, LNCS 5358 (Part I.), pp.772-781, 2008.

thinkoid / bs