抗混叠轮廓波域采用压缩感知的云图融合方法

提出一种采用压缩感知的云图融合方法.该方法针对传统轮廓波存在频谱混叠的缺点,结合抗混叠塔式滤波器组和方向滤波器组,构造出一种抗混叠的轮廓波变换,并将其引入压缩感知中的稀疏表示环节,将云图分解成稠密和稀疏两部分|对稠密成份采用传统方法进行融合,而对稀疏成份,则在压缩感知框架下,通过少数线性测量的融合,并采用二步迭代收缩的图像重构算法,在迭代时利用前面两个估计值更新当前值,得到融合结果.实验表明,该方法的融合结果无论在视觉质量及定量指标上都明显优于传统方法,有利于揭示全面的天气信息.

Cloud Image Fusion Using Compressed Sensing in Aliasing-free Contourlet Domain

A fusion method for meteorological cloud images based on compressed sensing (CS) was presented.Aiming at the frequency aliasing of the original contourlet transform,the aliasing-free pyramidal filter banks (AFPFB) was combined with directional filer banks (DFB) to construct a new transform,called aliasing-free contourlet transform (AFCT).Then,AFCT was applied to the sparse representation stage of CS to decompose the cloud image into dense and sparse components.The dense components were fused using conventional approach while the sparse components were fused under the framework of CS via fusing a few linear measurements by solving the two-step iterative shrinkage/threshold reconstruction algorithm which uses two previous estimates to obtain a new one.The experiment results demonstrate that the proposed method outperforms the traditional methods in terms of visual quality and quantitative criterion,and the fusion results is propitious to reveal the comprehensive weather information.