基于改进LDPC矩阵的遥感图像重构算法研究

Research on Remote Sensing Image Reconstruction Algorithm Based on Improved LDPC Matrix

针对当前遥感图像重构中存在的数据准确性和实效性差,以及传统的采样方法在图像采集上容易失真的问题,结合LDPC校验码提出一种改进的LDPC矩阵的图像重构方法。文章采用LDPC校验码的优势,取代传统置乱对角块矩阵中的对角块,以简化观测矩阵生成流程,进而构建了对角化的观测矩阵。最后,通过Lean图像和遥感图像为例,以PSNR和SSIM作为评价指标,对上述重构方法的效果进行验证。结果表明,重构后的Lean图像和遥感图像与其他的处理方法相比,都较为清晰,并且在PSNR和SSIM指标上,相对于其他的重构图像都具有优势。由此说明论文构建的对角化观测矩阵在图像重构方面,具有一定的优势和价值。

Aiming at the problem of poor data accuracy and effectiveness in current remote sensing image reconstruction and the distortion of traditional sampling methods in image acquisition,an improved image reconstruction method based on LDPC check code is proposed.In this paper,the advantage of LDPC check code is used to replace the diagonal block in the traditional scrambled diagonal block matrix,so as to simplify the process of generating observation matrix,and then the diagonalized observation matrix is constructed.Finally,taking Lean image and remote sensing image as examples,PSNR and SSIM are taken as evaluation indexes to verify the effectiveness of the above reconstruction methods.The results show that the reconstructed Lean image and remote sensing image are clearer than other processing methods,and have advantages over other reconstructed images in PSNR and SSM indexes.This paper shows that the diagonalization observation matrix constructed has certain advantages and value in image reconstruction.