摘要
遥感图像的退化主要受到噪声、模糊、混迭等多种因素的影响.针对单帧图像复原中的混迭现象,采用一种基于传感器几何特性的倒易晶胞复原方法.这种倒易晶胞的复原方法是把图像从空域对偶到频域,去除超出频域网格之间的频谱交叠部分,从而抑制混迭.然后,在频域采用逆滤波的方法进行去模糊和去彩色噪声.最后,利用复小波多方向可选择的特性进行去白噪声,可得到更准确的重建小波系数.实验表明,倒易晶胞结合复小波的复原方法比常用的复原方法效果更好.由于本复原算法考虑到了去混迭,因而比常用的图像复原方法更适于指导工程实践.
Remote sensing image is degraded by many factors, such as noise, blur and alias. A restoration of reciprocal cell is adopted to reduce alias of the single frame, which is based on geometrical characteristics of sensor. The restoration transform image into frequency domain. The superposition part of the frequency is deleted in order to dealias around the grids. And restoration of inverse filter is used to deblur and reduce the colour noise. Finally, method of the complex wavelet is combined to reduce the white noise. The reason is that the complex wavelet has property of the direction selecting. The test results indicate that the proposed restoration is more better than other conventional restorations. This restoration is fit to guiding practical projects on account of dealiasing.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2008年第10期1979-1985,共7页
Acta Electronica Sinica
基金
江苏省教育厅自然科学基金(No.BK2006704-2)
香港特区政府研究资助局资助(No.CUHK/4180/01E)
关键词
混迭
遥感图像
复原
复小波
倒易晶胞
彩色噪声
aliasing
remote sensing image
restoration
complex wavelet
reciprocal cell
colour noise
