研究目的
To develop a rigorous discretization model for the Spatial Spectral Compressive Spectral Imager (SSCSI) architecture, analyze the attainable spatial and spectral resolution, and demonstrate spectral zooming capabilities through simulations and experiments.
研究成果
The paper develops a rigorous discretization model for SSCSI, enabling spatial super-resolution and spectral zooming by adjusting the coded aperture position. Simulations show SSCSI outperforms CASSI and matches ideal colored CASSI in reconstruction quality. Experiments validate the model, demonstrating enhanced spectral resolution with increased s, though spatial quality degrades due to blurring. Future work includes optimizing coded aperture patterns and refining the model.
研究不足
The model assumes an infinitesimally small aperture in the objective lens, but in practice, finite aperture size introduces blurring effects, especially when s > 0, degrading spatial quality. Fabrication costs and light throughput are practical constraints. The spectral range may be limited if extreme wavelengths impinge outside the mask. The discretization model is an approximation and may not fully capture actual shearing effects.
1：Experimental Design and Method Selection:
The study involves a theoretical analysis and discretization of the SSCSI sensing process based on light propagation, followed by numerical simulations and experimental validation. The methodology includes deriving discrete measurement models for different scenarios of coded aperture and detector pitch sizes, and using compressive sensing algorithms for reconstruction.
2：Sample Selection and Data Sources:
A hyperspectral scene captured using a visible monochromator and CCD camera at the Computational Imaging and Spectroscopy Laboratory at University of Delaware is used as ground-truth in simulations. Experimental data are collected using a testbed setup with specific optical components.
3：List of Experimental Equipment and Materials:
Includes a TAMRON AF 70-300mm lens, transmissive diffraction grating (300 grooves/mm), 4f system with two 75mm 2" lenses, coded aperture mask, relay lens (35mm 1" lens), and a StingrayTM 640x480 CCD monochrome camera with 9.9μm pitch size. The coded aperture is printed on a photo-mask with pitch size of 19.4μm.
4：9μm pitch size. The coded aperture is printed on a photo-mask with pitch size of 4μm.
Experimental Procedures and Operational Workflow:
4. Experimental Procedures and Operational Workflow: The optical setup is assembled as described, with the coded aperture positioned at various distances (s parameters). Snapshots are captured for different coded aperture patterns, and the sensing matrix H is generated. Reconstruction is performed using the Gradient Projection for Sparse Reconstruction (GPSR) algorithm to solve an l1-minimization problem.
5：Data Analysis Methods:
Performance is evaluated using Peak Signal-to-Noise Ratio (PSNR) and correlation coefficients. The coherence of the sensing matrix is analyzed, and spectral resolution is determined by comparing coded patterns for adjacent bands.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
