现代电子技术
現代電子技術
현대전자기술
MODERN ELECTRONICS TECHNIQUE
2013年
14期
104-107
,共4页
高光谱分类%组合光谱特征%光谱导数特征%导数光谱技术
高光譜分類%組閤光譜特徵%光譜導數特徵%導數光譜技術
고광보분류%조합광보특정%광보도수특정%도수광보기술
hyperspectral classification%combinational spectral feature%spectrum derivative feature%derivative spectrum technology
在此分析了典型植被、岩矿、土壤和水体地物的光谱特性曲线及其一阶和二阶导数的特点与异同。研究表明通过光谱导数技术，可以快速分析出高光谱特性曲线的反射峰和吸收谷谱带的相关特征，且不同类别地物目标在原始光谱曲线和各阶光谱导数曲线上存在不同敏感波段。因此提出将高光谱原始数据和其各阶导数曲线的敏感波段综合起来构成组合光谱特征，以此达到优化高光谱特征性和降低分类难度，从而进一步提高分类性能的方法。实验结果表明组合光谱特征可以实现高光谱图像有效分类。
在此分析瞭典型植被、巖礦、土壤和水體地物的光譜特性麯線及其一階和二階導數的特點與異同。研究錶明通過光譜導數技術，可以快速分析齣高光譜特性麯線的反射峰和吸收穀譜帶的相關特徵，且不同類彆地物目標在原始光譜麯線和各階光譜導數麯線上存在不同敏感波段。因此提齣將高光譜原始數據和其各階導數麯線的敏感波段綜閤起來構成組閤光譜特徵，以此達到優化高光譜特徵性和降低分類難度，從而進一步提高分類性能的方法。實驗結果錶明組閤光譜特徵可以實現高光譜圖像有效分類。
재차분석료전형식피、암광、토양화수체지물적광보특성곡선급기일계화이계도수적특점여이동。연구표명통과광보도수기술，가이쾌속분석출고광보특성곡선적반사봉화흡수곡보대적상관특정，차불동유별지물목표재원시광보곡선화각계광보도수곡선상존재불동민감파단。인차제출장고광보원시수거화기각계도수곡선적민감파단종합기래구성조합광보특정，이차체도우화고광보특정성화강저분류난도，종이진일보제고분류성능적방법。실험결과표명조합광보특정가이실현고광보도상유효분류。
The spectral characteristic curves of typical vegetation,minerals,soil and water body,and similarities and dif?ferences of their first?order and second?order derivatives are analyzed. The research demonstrates that the relevant characteristics of the reflection peak and the absorption valley bands of hyperspectral spectral characteristic curves can be quickly obtained by analysis and spectrum derivative technology,and there are different sensitive bands between original spectral curve and various order spectral derivative curves of different categories objectives. Therefore,it is proposed that the composition of these sensitive bands of hyperspectral original data and each order derivative curve can constitute combinational spectral features to optimize the performance of hyperspectral features and reduce the classification difficulty. The experimental results show that this combi?national spectral features can realize the effective classification of hyperspectral images.
