双边滤波与矩匹配融合的高光谱影像条带噪声去除方法

A Strip Noise Removal Method of Hyperspectral Image Based on Combining Bilateral Filter with Moment Matching

针对环境减灾小卫星高光谱影像数据中存在的条带噪声,提出一种双边滤波与矩匹配方法融合的去除条带噪声方法。该方法克服了传统方法在去除条带噪声的同时带来新的噪声,并且边缘信息丢失等问题。最后进行了几种去除条带方法的比较试验,并且对试验结果进行了讨论。试验证明,提出的方法去除高光谱遥感影像的条带噪声效果明显,且能较好地保留原始影像中的地物信息。

Concerning the strip noise in hyperspectral image of HJ-1A satellite, a new strip noise removal method based on combining bilateral filter with moment matching is proposed in this article. The improved approach overcomes problems like new noise and loss of edge information with removing stripe noise in traditional methods. In the end of the paper, there are some comparative experiments between the improved algorithm and the original algorithms. The experimental results are discussed and can be seen that the strip noise removal method based on combining bilateral filter with moment matching has better effect for destriping of the hyperspectral image of HJ-1A satellite and preserves the more ground object information of original image.