A method of reconstructing compressive spectral imaging with a complementary prior constraint
Compressed spectral imaging (CSI) is a technique for acquiring a cube of spectral image data in a single snapshot. In this paper, we propose a reconstruction method for the CSI system that integrates complementary prior constraints on spectral data to significantly enhance the accuracy of reconstruc...
Gespeichert in:
| Veröffentlicht in: | Optics communications Jg. 550; S. 130010 |
|---|---|
| Hauptverfasser: | , , , , |
| Format: | Journal Article |
| Sprache: | Englisch |
| Veröffentlicht: |
Elsevier B.V
01.01.2024
|
| Schlagworte: | |
| ISSN: | 0030-4018, 1873-0310 |
| Online-Zugang: | Volltext |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Zusammenfassung: | Compressed spectral imaging (CSI) is a technique for acquiring a cube of spectral image data in a single snapshot. In this paper, we propose a reconstruction method for the CSI system that integrates complementary prior constraints on spectral data to significantly enhance the accuracy of reconstructed data. The fundamental concepts are as follows: (1) Firstly, by conducting a thorough analysis of the corresponding feature maps of spectral data on the convolution dictionaries, we confirm the feasibility of utilizing the stationary response characteristics of spectral data on the convolution dictionaries as a reconstruction constraint to enhance spectral accuracy. (2) To compensate for the limitations of Convolutional Sparse Coding in reconstructing low-frequency signals, this paper proposes utilizing iterative Total Variation operators to estimate the low-frequency portion of spectral data. This approach results in improved performance in reconstructing low-frequency data and noise suppression. (3) Convolutional Sparse Coding and Total Variation are utilized as constraints for high and low-frequency complementary reconstruction, resulting in a multi-constraint solution problem within the Plug-and-Play framework that simplifies the overall reconstruction process. The proposed method surpasses the state-of-the-art reconstruction methods in both spatial reconstruction quality and spectral accuracy, while also significantly enhancing the level of detail in reconstructed spectral images.
•Convolutional sparse coding shows stationary characteristics on spectral data.•Additional low frequency estimation is an important part of convolutional sparse coding.•Frequency complementary constraints can improve the accuracy of compressed sensing spectral imaging.•Iterative low frequency estimation can better estimate low frequency signal. |
|---|---|
| ISSN: | 0030-4018 1873-0310 |
| DOI: | 10.1016/j.optcom.2023.130010 |