结合NAPCA和复小波变换的高光谱遥感图像去噪

Denoising of hyperspectral remote sensing imagery using NAPCA and complex wavelet transform

提出了一种能够良好地保持高光谱遥感图像细节特征的噪声去除方法。该方法首先利用噪声调整的主成分分析（NAPCA）进行特征提取，再利用复小波变换（CWT）对NAPCA变换后的低能量成分进行去噪处理。对此低能量成分的每个波段利用二维复小波去噪，此时复小波系数采用BivaShrink函数进行收缩。然后对低能量成分的每条光谱进行一维复小波变换，利用邻域阈值函数进行小波系数的收缩。对AVIRIS 图像贾斯珀桥、月亮湖和盆地进行的仿真实验表明：该方法去噪后的信噪比与HSSNR相比提高了4.3-7.8 dB，与PCABS相比提高了0.8-0.9 dB，验证了该算法的可行性。真实数据OMIS图像的实验结果验证了该方法的有效性和适用性。

A new denoising algorithm was proposed to keep the fine features of hyperspectral remote sensing imagery effectively. Firstly, the noise-adjust principal components analysis （NAPCA） was performed on the hyperspectral datacube. Then output channels of the low-energy NAPCA were transformed into the wavelet domain by 2- D complex wavelet transform（CWT）. The BivaShrink function was used to shrink the wavelet coefficients. And then 1- D CWT denoising method was used to remove the noise of the each spectrum of the low-energy NAPCA datacube. The AVIRIS images Jasper Ridge, Lunar Lake and Low Altitude were used for the simulated experiment. Compared with the HSSNR and the PCABS, the signal-to-noise ratio （SNR） is improved by 4.3- 7.8 dB and 0.8- 0.9 dB via the proposed method in this paper, which shows that the proposed method is feasible. It is shown that the proposed method is correctable and available according to the experimental results of the real datacube OMIS.