<p>In the computer vision, background extraction is a promising technique. It is characterized by being applied in many different real time applications in diverse environments and with variety of challenges. Background extraction is the most popular technique employed in the domain of detecting moving foreground objects taken by stationary surveillance cameras. Achieving high performance is required with many perspectives and demands. Choosing the suitable background extraction model plays the major role in affecting the performance matrices of time, memory, and accuracy.</p><p>In this article we present an extensive review on background extraction in which we attempt to cover all the related topics. We list the four process stages of background extraction and we consider several well-known models starting with the conventional models and ending up with the state-of-the art models. This review also focuses on the model environments whether it is human activities, Nature or sport environments and illuminates on some of the real time applications where background extraction method is adopted. Many challenges are addressed in respect to environment, camera, foreground objects, background, and computation time. </p><p>In addition, this article provides handy tables containing different common datasets and libraries used in the field of background extraction experiments. Eventually, we illustrate the performance evaluation with a table of the set performance metrics to measure the robustness of the background extraction model against other models in terms of time, accurate performance and required memory.</p>

The background subtraction is a leading technique adopted for detecting the moving objects in video surveillance systems. Various background subtraction models have been applied to tackle different challenges in many surveillance environments. In this paper, we propose a model of pixel-based color-histogram and Fuzzy C-means (FCM) to obtain the background model using cosine similarity (CS) to measure the closeness between the current pixel and the background model and eventually determine the background and foreground pixel according to a tuned threshold. The performance of this model is benchmarked on CDnet2014 dynamic scenes dataset using statistical metrics. The results show a better performance against the state-of the art background subtraction models.

Abstract— Background subtraction is the dominant approach in the domain of moving object detection. Lots of research have been done to design or improve background subtraction models. However, there is a few well known and state of the art models which applied as a benchmark. Generally, these models are applied on different dataset benchmarks. Most of the time Choosing appropriate dataset is challenging due to the lack of datasets availability and the tedious process of creating the ground-truth frames for the sake of quantitative evaluation. Therefore, in this article we collected local video scenes for street and river taken by stationary camera focusing on dynamic background challenge. We presented a new technique for creating ground-truth frames using modelling, composing, tracking, and rendering each frame.  Eventually we applied nine promising benchmark algorithms used in this domain on our local dataset. Results obtained by quantitative evaluations exposed the effectiveness of our new technique for generating the ground-truth scenes to be benchmarked with the original scenes using number of statistical metrics. Furthermore, results shows the outperformance of SuBSENSE model against other tested models.

Background subtraction is the dominant approach in the domain of moving object detection. Lots of research has been done to design or improve background subtraction models. However, there are a few well-known and state-of-the-art models that can be applied as a benchmark. Generally, these models are applied to different dataset benchmarks. Most of the time, choosing an appropriate dataset is challenging due to the lack of dataset availability and the tedious process of creating ground-truth frames for the sake of quantitative evaluation. Therefore, in this article, we collected local video scenes of a street and river taken by a stationary camera, focusing on dynamic background challenges. We presented a new technique for creating ground-truth frames using modeling, composing, tracking, and rendering each frame. Eventually, we applied three promising algorithms used in this domain: GMM, KNN, and ViBe, to our local dataset. Results obtained by quantitative evaluations revealed the effectiveness of our new technique for generating the ground-truth scenes to be benchmarked with the original scenes using a number of statistical metrics.