研究目的
To propose a deterministic measurement matrix construction method for infrared image reconstruction using compressed sensing theory, addressing the limitations of random measurement matrices in storage and hardware implementation.
研究成果
The proposed deterministic measurement matrix based on the Archimedes spiral is feasible and outperforms Gaussian and Bernoulli random matrices in IR image reconstruction, providing higher PSNR and lower reconstruction error across various sampling rates, even for small targets and low sampling rates. This method offers advantages in storage efficiency and potential for hardware implementation.
研究不足
The study is based on simulations and does not involve real-world hardware implementation or testing with diverse IR image datasets. The performance is evaluated only against Gaussian and Bernoulli random matrices, and may not generalize to other types of matrices or noise conditions.
1：Experimental Design and Method Selection:
The methodology involves constructing a deterministic measurement matrix based on the Archimedes spiral equation. Points are sampled from the spiral to form a sequence, which is used to build the matrix. The sparse representation uses discrete cosine transform (DCT), and the reconstruction algorithm is basis pursuit (BP).
2：Sample Selection and Data Sources:
IR images of sizes 256x256 and 128x128 are used as test images, chosen for small and large target scenarios.
3：List of Experimental Equipment and Materials:
No specific equipment mentioned; simulations are conducted, likely using computational tools.
4：Experimental Procedures and Operational Workflow:
Define parameters (e.g., polar radius r, angle θ, coefficient b), sample points from the Archimedes spiral equation, construct the initial matrix, adjust for sparsity by setting elements >=1 to zero, and select rows based on sampling rate. Perform image reconstruction and compare with Gaussian and Bernoulli random matrices using peak signal-to-noise ratio (PSNR) and reconstruction error metrics.
5：Data Analysis Methods:
PSNR is calculated to evaluate image quality, and reconstruction error is analyzed. Comparisons are made across different sampling rates.
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
