ViBe: A Universal Background Subtraction Algorithm for Video Sequences

By Olivier Barnich, Marc Van Droogenbroeck. IEEE Transactions on Image Processing, 2011

1,940 citations recorded by the source corpus.

This paper presents a technique for motion detection that incorporates several innovative mechanisms. For example, our proposed technique stores, for each pixel, a set of values taken in the past at the same location or in the neighborhood. It then compares this set to the current pixel value in order to determine whether that pixel belongs to the background, and adapts the model by choosing randomly which values to substitute from the background model. This approach differs from those based upon the classical belief that the oldest values should be replaced first. Finally, when the pixel is found to be part of the background, its value is propagated into the background model of a neighboring pixel. We describe our method in full details (including pseudo-code and the parameter values used) and compare it to other background subtraction techniques. Efficiency figures show that our method outperforms recent and proven state-of-the-art methods in terms of both computation speed and detection rate. We also analyze the performance of a downscaled version of our algorithm to the absolute minimum of one comparison and one byte of memory per pixel. It appears that even such a simplified version of our algorithm performs better than mainstream techniques.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/3118c4280158c7b0ceffbb79c9f394dce51db5cc

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

The Visual Object Tracking VOT2015 Challenge Results (2015, methodology): “MKCF+ tracker is based on the multi-kernelized correlation filter tracker (MKCF) [62] and background modelling algorithm ViBe [5].”
SuBSENSE: A Universal Change Detection Method With Local Adaptive Sensitivity (2015, methodology): “Non-parametric models are also often considered for hardware [5] or high-speed parallel implementations due to their”
Deep background subtraction with scene-specific convolutional neural networks (2016, background): “In addition, it is almost exclusively based on low-level features such as color components ([1], [8]), gradients [9], or handcrafted features such as the LBP histogram [10], LBSP [7]”
Fast Object Segmentation in Unconstrained Video (2013, methodology): “Through extensive experiments on over 1400 video shots from two datasets [24, 19], we show that our method: (i) handles fast moving backgrounds and objects exhibiting a wide range of appearance, motions and deformations, including non-rigid and articulated objects; (ii) outperforms a state-of-the-ar”
Rain Removal in Traffic Surveillance: Does it Matter? (2018, methodology): “SuBSENSE is a method that builds on the spatial diffusion step introduced in ViBE [76].”
Background segmentation with feedback: The Pixel-Based Adaptive Segmenter (2012, methodology): “Therefore, we can observe a slight performance gain when adding gradient magnitudes.”
A Deep Convolutional Neural Network for Background Subtraction (2017, methodology): “[1] propose a sample based method for background modeling.”
Changedetection.net: A new change detection benchmark dataset (2012, methodology): “[15] and 3 stochastic methods based on background sample selection namely ViBe [2], ViBe+ [7], and Hofmann’s self-adaptive method (PBAS) [10].”
An innovative web-based collaborative platform for video annotation (2013, background): “For all these purposes, the scientific community has put a lot of effort in the development of powerful object detection [4, 13], tracking [8, 40] and recognition [30] approaches, which, however, are not able to scale up to many scenarios and objects due mainly to the lack of large labeled training ”
Saliency-Aware Video Object Segmentation (2018, methodology): “The segmentation methods in [1], [7], [8], [9], [22], [23], [27], [60], [61] and our method are unsupervised, while other methods in [59], [62] are supervised.”
Saliency-aware geodesic video object segmentation (2015, methodology): “The methods in [32, 23, 15, 6, 19, 4, 22] and our method are unsupervised.”
Detecting anomalous events in videos by learning deep representations of appearance and motion (2017, methodology): “ViBe has a low computational complexity and can obtain near real-time performance (almost 16 frames/second with a resolution of 360×240 in our Matlab environment).”
Fast human-animal detection from highly cluttered camera-trap images using joint background modeling and deep learning classification (2017, methodology): “ViBe [4] developed a non-parametric approach where the pixel-level background model is represented by a set of background samples.”
Deep Neural Network Concepts for Background Subtraction: A Systematic Review and Comparative Evaluation (2018, methodology): “Experimental results show that this method outperforms traditional background subtraction methods (MOG [180] and Codebook [209]) as well as recent state-of-art methods (ViBe [11], PBAS [81] and P2M [215]) on the CDnet 2012 dataset [62].”
Deep learning in video multi-object tracking: A survey (2019, methodology): “For example, CNNs have been exploited to reduce false positives in [72], where vehicle detections were obtained with a modified version of the ViBe algorithm [73] that performed background subtraction on the input.”
Interactive deep learning method for segmenting moving objects (2017, methodology): “Barnich and Van Droogenbroeck [3] proposed a method called “Vibe” whose per-pixel background model is made of a collection of N pixel values randomly selected over time.”