基于改进自适应流形的毫米波图像超分辨复原

PMMW Image Restoration Based on Improved Adaptive Manifolds

针对被动毫米波成像中图像分辨率低并伴有噪声的问题,本文提出了一种基于自适应流形高维滤波和高维空间中值滤波的被动毫米波复原算法。该方法利用限定窗口加权遍历法构建被动毫米波高维图像,然后利用主成分分析法提取高维图像的主要信息,通过高维的中值滤波去除降维后的被动毫米波图像冲激噪声和强噪声,最后通过自适应流形高维滤波对降噪后的图像进行滤波得到复原图像。仿真图像和实测图像的实验结果均表明,本文算法有效地增强了图像的分辨率,有利于被动毫米波图像的超分辨复原。

To solve the problem of poor resolution and noise in passive millimeter wave(PMMW) imaging,a PMMW restoration based on adaptive manifolds for high-dimensional filtering and high-dimensional median filtering is proposed in this paper.The high-dimensional PMMW image is generated by using weighted traversal algorithm,then the main information is extracted by PCA based dimension reduction,the retained data is by high-dimensional median filtering to remove the impulsive noise and strong noise,finally adaptive manifolds filter is applied to restore the image.The experiments with simulated and true PMMW images both show that proposed image is capable of enhancing the resolution