Method for compressed sensing features of spectrum

Abstract

The invention discloses a method for compressed sensing of the features of a spectrum, and the method can achieve the compressed extraction of a feature signal of a spectrum signal based on a constructed spectrum compression measurement matrix, and achieves the compressed sensing of the spectrum features. In addition, the method also proposes a physical implementation method for a spectrum compression measurement matrix Phi based on a DMD (digital micro-mirror device). The method can effectively avoid the signal reconstruction and feature extraction, and the measurement data can be directly used for classification and recognition, thereby greatly reducing the calculating and storage expenditure, and improving the obtaining efficiency of the spectrum information. Because the dimensions of aspectrum feature signal are smaller than the dimensions of a spectrum signal, the measurement data obtained through the method will be smaller than the measurement data of a conventional compressed spectrum imaging method.

Description

technical field [0001] The invention relates to a spectral information acquisition method, in particular to a compressed spectral feature sensing method. Background technique [0002] Spectral data can cover dozens or even hundreds of spectral frequency bands, and contains rich spectral features of targets, so it has important applications in the fields of target detection, classification, and identification. However, the spatial spectrum dimension of spectral data is high, which causes great difficulties in the acquisition, storage, and processing of spectral data, and at the same time poses a huge challenge to the real-time performance of spectral feature extraction and classification recognition. Compressive sensing theory provides a new idea for solving the dimension disaster of spectral data, which is expected to improve the efficiency of spectral data acquisition and processing. Compressed sensing theory shows that when the signal is sparse, the original signal can be...

AI Technical Summary

Problems solved by technology

[0004] Feature reconstruction can avoid subsequent operations such as spectral information processing and feature extraction after reconstruction, but the key link of sparse reconstruction cannot be avoided, so reconstruction also requires calculation, time and other overhead

Images