Online Tensor Robust Principal Component Analysis

Online robust principal component analysis (RPCA) algorithms recursively decompose incoming data into low-rank and sparse components. However, they operate on vectors cannot directly be applied to higher-order arrays (e.g. video frames). In this paper, we propose a new online PCA algorithm that preserves the multi-dimensional structure of data. Our is based recently proposed tensor singular value decomposition (T-SVD). We develop convex optimization-based approach recover component; subsequently, update using incremental T-SVD. an efficient convolutional extension fast iterative shrinkage thresholding (FISTA) produce solve optimization problem. demonstrate tensor-RPCA with application background foreground separation in stream. The modeled as signal. gradually changing subspace. Extensive experiments real-world videos are presented results effectiveness our PCA.