An online tensor robust PCA algorithm for sequential 2D data

Tensor robust principal component analysis (PCA) approaches have drawn considerable interests in many applications such as background subtraction, denoising, and outlier detection, etc. In this paper we propose an online tensor robust PCA where the multidimensional data (tensor) is revealed sequentially in online mode, and tensor PCA is updated based on the latest estimation and the newly collected data. Compared to the tensor robust PCA in batch mode, we significantly reduce the required memory and improve the computation efficiency. Application on fusing cloud-contaminated satellite images demonstrates that the proposed method shows superiority in both convergence speed and performance compared to the state-of-the-art approaches.