光谱相似性度量方法研究进展

发布时间：2018-04-21 01:00

本文选题：高光谱图像 + 光谱相似性度量　；参考：《哈尔滨工程大学学报》2017年08期

【摘要】：为了进一步分析光谱相似性度量在高光谱图像处理中的应用,从距离、投影等角度充分归纳总结了现有二元光谱相似度量方法,并分析讨论了二元光谱相似度量存在的问题。重点介绍了一种多元光谱相似性测量方法,也称N维立体光谱角(N-dimensional solid spectral angle,NSSA)方法。NSSA方法从本质上突破了传统的二元光谱角(spectral angle mapping,SAM)仅能计算两个光谱之间夹角的局限性,具备联合计算多元光谱欧氏空间夹角的能力,为评价多元光谱联合相似性提供了一种定量化的度量手段。最后,对NSSA方法在高光谱波段选择及端元提取领域的潜在研究价值和应用现状进行了分析和展望。通过分析表明NSSA方法所具备的特性可更好地实现光谱相似性度量,在高光谱图像处理领域具有较高的研究价值。
[Abstract]:In order to further analyze the application of spectral similarity measurement in hyperspectral image processing, the existing binary spectral similarity methods are summarized from the angles of distance and projection, and the existing problems of binary spectral similarity are analyzed and discussed. A multivariate spectral similarity measurement method, also known as N-dimensional solid spectral angleNSSA. NSSA, is introduced in this paper, which essentially breaks through the limitation that the traditional binary spectral spectrum angle mapping SAM can only calculate the angle between the two spectra. It has the ability to jointly calculate the Euclidean space angle of multivariate spectra, which provides a quantitative measure for evaluating the joint similarity of multivariate spectra. Finally, the potential research value and application status of NSSA method in the field of hyperspectral band selection and end-component extraction are analyzed and prospected. The analysis shows that the characteristics of the NSSA method can better measure the spectral similarity and have high research value in the field of hyperspectral image processing.
【作者单位】：哈尔滨工程大学信息与通信工程学院;
【基金】：国家自然科学基金项目(61405041,61571145) 黑龙江省自然科学基金重点项目(ZD201216) 哈尔滨市优秀学科带头人基金项目(RC2013XK009003)
【分类号】：TP751

【相似文献】