基于低秩约束和边信息的近似消息传递CS重构算法
发布时间：2018-12-19 11:41
【摘要】：噪声环境下图像压缩感知(compressive sensing,CS)重构方法的性能会大幅度下降。在近似消息传递(approximate message passing,AMP)算法的基础上,同时利用结构先验信息和边信息来增强AMP算法对噪声的鲁棒性。利用图像中相似块的低秩特性,在反投影的含噪图像中捕获低秩子空间的结构特征;再将含有确定成分的前期重构图像作为边信息,以实现细节的增强。实验表明,本文算法比原始AMP算法在峰值信噪比(peak signal to noise ratio,PSNR)上平均提高了3.89dB,且获得更加清晰的重构图像;与仅利用低秩特性的AMP算法相比,引入边信息后本文算法在PSNR上获得了0.27dB的增益,同时增强了重构图像的细节。
[Abstract]:The performance of image compression sensing (compressive sensing,CS) reconstruction in noisy environment will be greatly reduced. Based on the approximate message passing (approximate message passing,AMP (approximate message passing,AMP) algorithm, the structural prior information and edge information are used to enhance the robustness of the AMP algorithm to noise. By using the low rank characteristic of similar blocks in the image, the structural features of the low rank subspace are captured in the noisy image with backprojection, and the pre-reconstructed image with deterministic components is used as edge information to enhance the details. Experimental results show that the proposed algorithm is 3.89 dB higher than the original AMP algorithm on the peak signal-to-noise ratio (peak signal to noise ratio,PSNR), and a clearer reconstructed image is obtained. Compared with the AMP algorithm which only uses the low rank characteristic, the 0.27dB gain is obtained on the PSNR by introducing edge information, and the details of the reconstructed image are enhanced at the same time.
【作者单位】： 华南理工大学电子与信息学院;
【基金】：国家自然科学基金(61471173)资助课题
【分类号】：TP391.41
本文编号：2386872
[Abstract]:The performance of image compression sensing (compressive sensing,CS) reconstruction in noisy environment will be greatly reduced. Based on the approximate message passing (approximate message passing,AMP (approximate message passing,AMP) algorithm, the structural prior information and edge information are used to enhance the robustness of the AMP algorithm to noise. By using the low rank characteristic of similar blocks in the image, the structural features of the low rank subspace are captured in the noisy image with backprojection, and the pre-reconstructed image with deterministic components is used as edge information to enhance the details. Experimental results show that the proposed algorithm is 3.89 dB higher than the original AMP algorithm on the peak signal-to-noise ratio (peak signal to noise ratio,PSNR), and a clearer reconstructed image is obtained. Compared with the AMP algorithm which only uses the low rank characteristic, the 0.27dB gain is obtained on the PSNR by introducing edge information, and the details of the reconstructed image are enhanced at the same time.
【作者单位】： 华南理工大学电子与信息学院;
【基金】：国家自然科学基金(61471173)资助课题
【分类号】：TP391.41
【相似文献】
相关期刊论文 前2条
1 王轶冰;;学习原型超平面融合SILD的户外人脸识别[J];计算机工程与应用;2014年10期
2 ;[J];;年期
相关硕士学位论文 前1条
1 刘瑞娜;融入边信息的多变量IB方法的多视角聚类算法研究[D];郑州大学;2014年
，本文编号：2386872
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2386872.html
