论文部分内容阅读
针对遥感卫星成像高像素多谱段而星上资源有限的特点,将对自然图像的分块压缩感知及重建算法拓展到多光谱遥感,提出了一种优化预测残差的多光谱图像重建算法.通过优化SPL迭代重建算法和最新CCSDS-MHDC多光谱无损压缩算法的预测器,在测量域2范数约束下以迭代的方式提高预测残差压缩感知的重建效果和算法效率.实验结果表明:相比其他单波段独立重建的算法,该算法在显著提高图像重建主客观质量的同时,减少了计算复杂度和压缩采样端和重建端的内存使用量,提高了大视场遥感图像感知重建的可实现性.
Aiming at the characteristics of high pixel multi-spectral images and limited on-satellite resources for remote sensing satellite imagery, this paper extends the block compression perception and reconstruction algorithm of natural images to multispectral remote sensing, and proposes a multispectral image reconstruction algorithm to optimize prediction residuals. By optimizing the SPL iterative reconstruction algorithm and the latest predictor of the CCSDS-MHDC multispectral lossless compression algorithm, the reconstruction effect and algorithm efficiency of the prediction residual compression perception are improved iteratively under the constraints of the measurement domain 2 norm. The experimental results show that: Compared with other single-band independent reconstruction algorithms, the proposed algorithm not only improves the subjective and objective quality of image reconstruction significantly but also reduces the computational complexity and memory usage of compressed sampling and reconstruction ends, and improves the realization of perceptual reconstruction of large-field remote sensing images Sex.