Foreground Detection Using the Choquet Integral

Abstract

Foreground Detection is a key step in background subtraction problem. This approach consists in the detection of moving objects from static cameras through a classification process of pixels as foreground or background. The presence of some critical situations i.e noise, illumination changes and structural background changes produces an uncertainty in the classification of image pixels which can generate false detections. In this context, we propose a fuzzy approach using the Choquet integral to avoid the uncertainty in the classification. The experiments on different video datasets have been realized by testing different color space and by fusing color and texture features. The proposed method is characterized through robustness against illumination changes, shadows and little background changes, and it is validated with the experimental results.

A hybrid framework combining background subtraction and deep neural networks for rapid person detection

“…Zhang and Xu [35] performed background subtraction by using a similarity criterion, which has the characteristics of color and texture of the input image. El Baf et al [36] obtained satisfactory results by using the Choquet Integral, which has the characteristics of color, edge and texture.…”

Deep detector classifier (DeepDC) for moving objects segmentation and classification in video surveillance

“…This survey is composed of 50 articles but covered only basic models based on fuzzy concepts. El Baf et al [21] classified pixels as belonging to the foreground or the background to detect moving objects from static cameras based on a fuzzy approach using the Choquet integral to avoid classification uncertainty. Experiments on different video datasets were performed by testing different colour spaces and fusing the colour and texture characteristics.…”

New technique for online object tracking‐by‐detection in video

“…We used this extended texture‐based technique, along with the Jabri's colour and edge‐based technique [19], to obtain the benefits of colour, edge and texture features. Then, Choquet integral [36] is used in background subtraction to aggregate the confidence maps extracted using the corresponding operators. One PF is then employed per each detected target using a small state space to make use of the features calculated continuously during the detection phase along with their corresponding confidence maps.…”

“…A simple model was developed to make use of the generative model to consider the appearance change and the discriminative boosting classifier to separate the foreground target effectively from the background, achieving a more flexible and robust likelihood function for particle filter (stochastic search mechanism). The new proposed technique used Choquet fuzzy integral [36, 37] in the classification step, since the scale is continuum. In other words, it is constructed as a cardinal one, where the distances or the differences between the values can be defined.…”