Signal compression and recovery method and system based on tensor decomposition and deep learning

Abstract

The invention discloses a signal compression and recovery method and system based on tensor decomposition and deep learning, and the method comprises the following steps: generating a measurement matrix according to a preset signal sampling rate; multiplying the original signal by the measurement matrix to generate a compressed signal; establishing a neural network based on a tensor decompositionmethod for signal recovery, decomposing the neural network through the tensor decomposition method to obtain a test signal which tends to an original signal, and obtaining the trained neural network by using the compressed signal and the original signal; and multiplying the test signal by the measurement matrix to obtain a compression test signal, and inputting the compression test signal into thetrained neural network to obtain a recovered test signal. According to the method, the time required by signal recovery can be greatly shortened, meanwhile, the number of parameters required by the network can be greatly reduced, the required calculation space is reduced, and high signal recovery precision can still be kept even when the signal acquisition rate is low.

Description

technical field [0001] The present invention relates to the technical field of signal processing, in particular to a signal compression and restoration method and system based on tensor decomposition and deep learning. Background technique [0002] As people have higher and higher resolution requirements for multimedia content such as images and videos, the amount of data sampled based on the Nyquist sampling law is too large, which is not conducive to storage and transmission, and there are many redundancies in the data itself. , can be further compressed, so the compressed sensing (Compressed sensing, CS) technology is proposed. Compressed sensing is a method for acquiring and reconstructing sparse or compressible signals, which uses the sparse characteristics of the signal, compared with the Nyquist theory, to restore the original desired information from fewer measurements. signal of. However, there are two main problems in the traditional compressed sensing method: fi...

AI Technical Summary

Problems solved by technology

However, there are two main problems in the traditional compressed sensing method: first, many actual signals are not completely sparse on some base classes, which affects the accuracy of signal recovery; second, the traditional compressed sensing algorithm converges slowly, which limits The application field of compressive sensing

Although this method can solve the problem of high signal dimensionality, it often leads to blocking artifacts in the recombined large signal, especially when the signal acquisition rate is relatively low

Images