The Hyperspectral images have the two-dimensional spatial information data and one-dimensional spectral information data,information between spectrum has strong correlation. Aiming at the characteristic of hyperspectral image,a hyperspectral imagery denoising method based on two dimensional empirical mode decomposition is proposed. At first,each band of the hyperspectral image is decomposed respectively by two dimensional empirical mode,and different scales of intrinsic mode functions are obtained. According to spectral correlation between the bands containing seriously noise and the bands containing low noise,the weight values are calculated,The high frequency intrinsic mode functions coefficients of the bands containing low noise are weighted and summed. The original intrinsic mode functions coefficients of the band containing seriously noise are replaced by the sum of weighted high intrinsic mode functions coefficients of the band which contains weak noise. Finally,the denoised hyperspectral images are achieved by inverse two dimensional empirical mode decomposition. The experimental results show that the proposed method can remove the noise of hyperspectral image effectively and keep the detail well. Comparing with the classical wavelet denosing method,the denoised images by the proposed method have a higher SNR and better visual effects.
参考文献
相似文献
引证文献
引用本文
厉祥,王文波.基于二维经验模态分解的高光谱影像去噪方法[J].激光与红外,2013,43(11):1311~1315 LI Xiang, WANG Wen-bo. Hyperspectral image denoising based on bidimensional empirical mode decomposition[J]. LASER & INFRARED,2013,43(11):1311~1315