论文部分内容阅读
地物光谱特征分析是对地物进行分类和匹配的基础,目前高光谱遥感技术应用在精细物种识别中主要采用波谱分析的方法。重点探索非线性空间里的相似性测度方法,由于光谱曲线表征复杂光谱成像的非线性过程,论文从空间目标的整体形状描述非空集合之间的差异,采用Fr′echet距离、Hausdorff距离、Euclidian距离分别定义光谱特征曲线的距离,设计算法测量光谱向量之间的非线性相似程度。结果表明,采用Fr′echet距离、Hausdorff距离、Euclidian距离度量光谱相似度的精度依次减弱,但依据Fr′echet距离的算法时间复杂度略高。基于Fréchet距离的方法充分考虑了曲线上点的位置信息及整体曲线的走势问题,其在精度、抗噪能力等方面均有提升,从而为分析光谱特征提供了可能的新途径。
The spectral feature analysis of ground objects is the basis for the classification and matching of ground objects. At present, the application of hyperspectral remote sensing technology in fine species identification mainly uses the method of spectral analysis. This paper focuses on the similarity measure method in nonlinear space. Since the spectral curve characterizes the nonlinear process of complex spectral imaging, the thesis describes the differences between non-empty sets from the overall shape of the spatial target. The Fréchet distance, Hausdorff distance, Euclidian Distances Define the distance of the spectral characteristic curve respectively, and the design algorithm measures the degree of non-linear similarity between the spectral vectors. The results show that the accuracy of the spectral similarity of the measures of Fr’echet distance, Hausdorff distance and Euclidian distance decreases in turn, but the algorithm based on Fr’echet distance has a slightly higher time complexity. The method based on Fréchet distance fully considers the position information of points on the curve and the tendency of the whole curve, which improves the accuracy and anti-noise ability, thus providing a new possible way to analyze the spectral characteristics.