论文部分内容阅读
超谱遥感图像包含了大量的波段,波段之间的相关性较高,采用信息融合技术可以降低超谱图像的分析难度。提出了一种结构新颖的第二代小波加权融合算法。首先将图像分解为两个序列,用2阶Neville滤波器构造预测和更新算子,对两个序列以矩形栅格和梅花形栅格的格式进行交替预测和更新;再以各个波段的方差作为融合的特征,进行特征级第二代小波加权融合,最后对图像进行第二代小波重构。为了验证新方法的有效性,采用机载可见光红外成像光谱仪超谱遥感图像进行仿真,并与典型融合方法主成分分析和离散小波变换的融合效果相比较。实验结果表明提出的第二代小波加权融合算法能够很好地保持图像的空间特性和光谱特性,其熵值高于主成分分析融合结果0.1949,高于离散小波变换融合结果0.7998。
Hyperspectral remote sensing image contains a large number of bands, the correlation between the bands is high, the use of information fusion technology can reduce the difficulty of hyperspectral image analysis. A novel second-generation wavelet weighted fusion algorithm is proposed. Firstly, the image is decomposed into two sequences, the prediction and update operator is constructed by using 2-order Neville filter, and the two sequences are alternately predicted and updated in the format of rectangular grid and quincunx grid. Then the variance of each band is taken as Fusion characteristics of the characteristics of the second generation of wavelet-weighted fusion, and finally the image of the second generation of wavelet reconstruction. In order to verify the effectiveness of the new method, the hyperspectral remote sensing image of airborne visible-light infrared imaging spectrometer was used to simulate the fusion effect of the proposed method and the fusion effect of the principal component analysis and discrete wavelet transform of the typical fusion method. The experimental results show that the proposed second-generation wavelet weighted fusion algorithm can well preserve the spatial and spectral characteristics of the image, and its entropy value is higher than the fusion result of principal component analysis 0.1949, which is higher than the fusion result of discrete wavelet transform 0.7998.