基于空谱联合和波段分类的高光谱压缩感知重构

卫星在获取地面信息时会受到大气、电磁波的干扰,导致高光谱影像本身产生坏线和噪声。针对这一问题,本文结合高光谱遥感影像的特性提出了一种基于空谱联合和波段分类的影像重构方法。首先,根据噪声影响程度将影像波段分为坏线强干扰波段和非干扰低噪声波段;其次,对波段进行分组,确定每组参考波段,并对参考波段进行独立重构;然后,根据参考波段构建双模式空谱联合预测模型,利用正则化交叉投影得到非参考波段重构影像;最后,对坏线强干扰波段,先进行独立重构,然后对重构影像进行小波分解,通过高频校正得到了干扰波段最终重构影像。实验表明,本文方法对重构高光谱影像的平均信噪比较传统方法提高了1~2dB。

Hyperspectral compressed perceptual reconstruction based on space spectrum combination and band classification

Satellites are subject to the interference of the atmosphere and electromagnetic wave when obtaining ground information, resulting in the occurrence of bad lines and noise in the hyperspectral image itself. In response to this problem, this paper, taking the characteristics of hyperspectral remote sensing images into account, presents an image reconstruction method based on the combination of hyperspectral and band classification. Firstly, according to the level of noise, the image band is classified as two levels:strong interference band due to bad line and the non-interference low noise band. Secondly, the second level is divided into groups, and then the standard band is determined. The standard band is independently reconstructed and the dual-mode space spectrum joint prediction model is constructed referring to the standard band. The reconstruction image of non-reference band is obtained by cross-projection using the regularization theory. Finally, the strong interference band is reconstructed independently and the reconstructed image is decomposed by wavelet transform, leading to the final image of the interference band by high-frequency correction. Experimental results show that the proposed method improves the average signal-to-noise ratio of the reconstructed image by about 1~2 dB compared with the traditional one.