Incremental and Multi-feature Tensor Subspace Learning

Introduction

Background subtraction is the art of separating moving objects from their background. The background modeling is one of the main steps of the BS process. Several subspace learning algorithms based on matrix and tensor tools have been used to perform the background modeling of the scenes. However, several subspace learning algorithms work on a batch process increasing memory consumption when data size is very large. Moreover, these algorithms are not suitable for streaming data when the full size of the data is unknown. In this work, we propose an incremental tensor subspace learning that uses only a small part of the entire data and updates the low-rank model incrementally when new data arrive. In addition, the multi-feature model allows us to build a robust low-rank background model of the scene. Experimental results shows that the proposed method achieves interesting results for background subtraction task.

Principle

Figure 1 shows the block diagram of the proposed approach. In the step (a), the last N frames from a streaming video are stored in a sliding block or tensor At. Next, a feature extraction process is done at step (b) and the tensor At is transformed in another tensor (step (c)) . In (d), an incremental higher-order singular value decomposition (iHoSVD) is applied in the tensor resulting in a low-rank tensor. Finally, in the step (e) a foreground detection method is applied for each new frame to segment the moving objects.

Figure 1: Incremental and Multi-feature Tensor Subspace Learning

Publication

A. Sobral, C. Baker, T. Bouwmans, E. Zahzah, “Incremental and Multi-feature Tensor Subspace Learning applied for Background Modeling and Subtraction”, International Conference on Image Analysis and Recognition, ICIAR 2014, October 2014. [pdf]