Video reconstruction using inductive three dimensional sparsity measure

Compressive sensing (CS) aims at acquiring and reconstructing sparse signals at a low sampling rate, and thus has wide applications in video processing. In this paper, an inductive three-dimensional sparsity measure (I_3DSM) is proposed for real-time video reconstruction. In the proposed sparsity measure, we utilize an online trained projection matrix to exploit the low-rank property of video sequence in the sparse transform domain. A large number of experiments are conducted to illustrate the superior performance of I_3DSM as compared with some known sparsity measures.