论文部分内容阅读
目的针对光谱角制图(SAM)分类算法对高光谱像元光谱曲线的局部特征和其辐射强度不敏感,而且易受噪声和维数灾难影响,致使分类效率低和精度较差等缺陷,将谐波分析(HA)技术引入到SAM高光谱影像分类中,提出一种基于谐波分析的光谱角制图(HA-SAM)高光谱影像分类算法。方法利用HA技术将高光谱影像从光谱维变换到能量谱特征维空间,并提取低次谐波分量及特征系数(谐波余项、相位和振幅),用特征系数组成的向量代替光谱向量,对高光谱影像进行SAM分类。结果将SAM和HA-SAM同时应用于EO-1卫星的Hyperion高光谱影像分类,通过对比和分析,验证了HA-SAM的优越性,再选择AVIRIS(airborne visible infrared imaging spectrometer)高光谱影像对HA-SAM进行验证,结果表明该算法具有较强的普适性。结论 HA-SAM提高了传统SAM高光谱影像分类的效率和精度,而且适用性较强具有良好的应用前景。
Aim To solve the problem that spectral classification (SAM) classification algorithm is not sensitive to the local characteristics and radiation intensity of hyperspectral spectral curve and is easily affected by noise and dimensionality disaster, the classification efficiency is low and the accuracy is poor. The wave analysis (HA) technique is introduced into the SAM hyperspectral image classification, and a hyperspectral imaging (HA-SAM) hyperspectral image classification algorithm based on harmonic analysis is proposed. Methods The hyperspectral image was transformed from spectral dimension to energy spectral feature space by using HA technique. The low-order harmonic components and characteristic coefficients (harmonic remainder, phase and amplitude) were extracted, and the spectral vectors were replaced by vectors composed of characteristic coefficients. SAM classification of hyperspectral images. Results The SAM and HA-SAM were simultaneously applied to the Hyperion hyperspectral image classification of EO-1 satellite. The superiority of HA-SAM was verified by comparison and analysis. Then, the airborne visible infrared imaging spectrometer (AVIRIS) -SAM, the results show that the algorithm has strong universality. Conclusion HA-SAM improves the efficiency and accuracy of the traditional SAM hyperspectral image classification, and has good applicability as a strong applicability.