Journals

T. Bouwmans, Special Issue "Detection of Moving Objects", MDPI Journal of Imaging, to appear in 2018

The detection of moving objects is one of the most important steps in the video processing field, such as in video-surveillance, optical motion capture, multimedia applications, teleconferencing, video editing, human-computer interface, etc. The last two decades have witnessed very significant publications on the detection of moving objects in video taken by static cameras; however, recently, new applications in which backgrounds are not static, such as recordings taken from drones, UAVs or Internet videos, need new developments to detect robustly moving objects in challenging environments. Thus, effective methods for robustness to deal both with dynamic backgrounds and illumination changes in real scenes with fixed cameras or mobile devices are needed and so different models need to be used such as advanced statistical models, fuzzy models, robust subspace learning models and deep learning models.

The intent of this special issue is to provide: 1) new approaches in detection of moving objects, 2) new strategies to improve foreground detection algorithms to tackle critical scenarios, such as dynamic backgrounds, illumination changes, night videos and low-frame rate videos, and 3) new adaptive and incremental algorithms to achieve real-time applications. (more information)

A. Petrosino, L. Maddalena, T. Bouwmans, Special Issue on "Scene Background Modeling and Initialization", Pattern Recognition Letters, September 2017.

This special issue on scene background modeling and initialization has collected original papers with innovative contributions to the research and the development related to theory, methods, algorithms, evaluation, and applications of the considered subject. Eleven papers have been accepted for publication by taking into account the technical quality, the originality, and the innovation of the presented ideas, solutions, and applications. (more information)

T. Bouwmans, L. Davis, J. Gonzalez, M. Piccardi, C. Shan, Special Issue on "Background Modeling for Foreground Detection in real-world dynamic scenes”, Machine Vision and Applications, MVA 2014, July 2014.

Background modeling and foreground detection are important steps in the video processing field such as video-surveillance, optical motion capture, multimedia applications,teleconferencing, video editing, human-computer interface, etc. Conventional foreground detection exploits change detection in video sequences. Some algorithms have used frame difference, but the most common approach is background subtraction that detects moving objects or abandoned objects in video sequences taken from a fixed camera. The last decade witnessed very significant publications on background subtraction but recently new applications in which background is not static, such as recordings taken from mobile devices or Internet videos, need new developments to detect robustly moving objects in challenging environments. Thus, effective methods for robustness to deal both with dynamic backgrounds, illumination changes in real scene with fixed cameras or mobile devices are needed and so different strategies may be used such as automatic feature selection, model selection or hierarchical model. Another feature of background modeling methods is that the use of advanced models has to be computed in real-time and low memory requirements. Algorithms may need to be redesigned to meet these requirements.

The goals of this special issue are threefold: 1) proposing new mathematical tools in background modeling and foreground detection, 2) developing new strategies to improve foreground detection algorithms to tackle critical situations such as dynamic backgrounds and illumination changes and 3) developing new adaptive and incremental algorithms to achieve real-time applications. (more information)

A. Vacavant, L. Tougne, T. Chateau, Special section on "Background models comparison", Computer Vision and Image Understanding, CVIU 2014, May 2014.

Background subtraction consists in detecting moving objects in video sequences, for various kinds of computer vision applications: video-surveillance, gesture analysis, motion estimation, video compression, etc. The methods developed for this purpose aim at modeling and learning the background of the scene, to deduce consequently the foreground by subtracting processed input frames from this learned background.

Although the evaluation of background subtraction techniques is an important issue, the impact of relevant papers that handle both benchmarks and annotated dataset is limited. Moreover, a majority of authors that propose a novel approach compare their work with classical techniques rwith a restricted part of the literature, but rarely with numerous recent related works. Recently, this lack of durable reference has led to the emergence of several benchmarks, fully available on the Web, as ChangeDetection.net or SABS (Stuttgart Artificial Background Subtraction Dataset). These datasets allow authors to download challenging videos, and to compare their work with both classical and recent contributions.

The Background Models Challenge (BMC) is a benchmark proposing a set of both synthetic and real videos, together with several complementary performance evaluation criteria. The first workshop BMC, organized within the ACCV (Asian Conference on Computer Vision) 2012 has been mainly dedicated to evaluate and rank background/foreground algorithms for intelligent video-surveillance applications. In the continuity of this workshop, this special issue “Background Models Comparison” aims at exposing new original, competitive methodologies for BS, and at evaluating them thanks to the benchmark. (more information)