Compressed video capture and reconstruction system based on data drive tensor subspace

Abstract

The invention provides a compressed video capture and reconstruction system based on a data drive tensor subspace. The compressed video capture and reconstruction system comprises a tensor sparse base structure module, a video signal sensing module and a reconstruction processing module, wherein the tensor sparse base structure module utilizes a tensor subspace learning method to generate a sparse base matrix corresponding to the tensor subspace, the video signal sensing module projects a video signal in a tensor block mode to obtain an observed value, and the reconstruction processing module receives the sparse base matrix and the observed value and performing decoding reconstruction on all dimensionalities of a tensor signal respectively. The compressed video capture and reconstruction system provides compressed sampling, meanwhile conforms to a distributed progressive structure in the video sampling process, and also improves the reconstruction accuracy and efficiency of the special structure of the tensor sparse base matrix. The compressed video capture and reconstruction system greatly improves the video signal sampling efficiency, obtains reconstruction gain at different sampling compression rates compared with other methods, and meanwhile has good expandability.

Description

technical field [0001] The invention relates to a video signal acquisition scheme, in particular to a compressed video acquisition and reconstruction system based on data-driven tensor quantum space. Background technique [0002] The acquisition and encoding (compression) of video signals is crucial for applications such as storage and transmission of video. The traditional signal processing system adopts the mode of sampling first and then compressing: in order to completely preserve all information of the signal, the video should be sampled at a sampling frequency not less than twice the signal bandwidth; the collected original signal is removed after a series of encoding techniques For the purpose of redundancy, the bottleneck of related technologies is that a large amount of sensors and computing resources are spent to obtain a small amount of signal compression data after processing, and the demand for resources at the sampling end is too high. In order to further impr...

AI Technical Summary

Problems solved by technology

[0003] After searching the literature of the prior art, it was found that in the article "Generalized tensor compressive sensing" published by Q.Li, D.Schonfeld and S.Friedland at the "IEEE International Conference on Multimedia and Expo" (IEEE ICME) conference in 2013 A reconstruction based on the discrete cosine transform (DCT) tensor basis is proposed to apply compressed sensing to video sampling. This method directly compresses and samples the video tensor at the sampling encoding end using a sensing matrix for each dimension, and at the decoding end uses The DCT base is used as a sparse base to reconstruct the signal. This method can effectively improve the efficiency of video sampling and ensure the subjective quality of the reconstructed video. However, the DCT base used in this method is a fixed base. For For video scenes with complex textures or severe motion, the DCT base used in this method is not flexible enough to accurately and effectively sparsely represent the tensor blocks of video frames, and cannot provide more effective sparsity and adaptability, resulting in reduced effects

Images