信号处理
信號處理
신호처리
SIGNAL PROCESSING
2015年
4期
414-422
,共9页
高光谱图像%负相似%LapSVM算法%LNP 算法%半监督分类
高光譜圖像%負相似%LapSVM算法%LNP 算法%半鑑督分類
고광보도상%부상사%LapSVM산법%LNP 산법%반감독분류
hyperspectral image%dissimilarity%LapSVM%linear neighborhood propagation (LNP)%semisupervised clas-sification
高光谱图像数据体现为波段多、地物标签获取困难大、谱信息抗干扰能力弱等特征,容易引起维数灾难、光谱空间变异性等问题,从而影响分类器的分类精度。针对这些问题,本文将负相似信息引入到拉普拉斯支持向量机(Laplacian Support Vector Machine,LapSVM)的流形正则化项中,提出了一种引入负相似的拉普拉斯支持向量机(Dissimilarity in Laplacian Support Vector Machine,Diss-LapSVM)分类算法,抑制光谱空间变异对分类结果的影响。同时,本文提出利用线性近邻传播(Linear Neighborhood Propagation,LNP)算法构造图的拉普拉斯矩阵,更有效地引入无标签样本的信息。实验结果表明,本文算法的分类精度得到了提高,特别是对光谱特征相似的地物。
高光譜圖像數據體現為波段多、地物標籤穫取睏難大、譜信息抗榦擾能力弱等特徵,容易引起維數災難、光譜空間變異性等問題,從而影響分類器的分類精度。針對這些問題,本文將負相似信息引入到拉普拉斯支持嚮量機(Laplacian Support Vector Machine,LapSVM)的流形正則化項中,提齣瞭一種引入負相似的拉普拉斯支持嚮量機(Dissimilarity in Laplacian Support Vector Machine,Diss-LapSVM)分類算法,抑製光譜空間變異對分類結果的影響。同時,本文提齣利用線性近鄰傳播(Linear Neighborhood Propagation,LNP)算法構造圖的拉普拉斯矩陣,更有效地引入無標籤樣本的信息。實驗結果錶明,本文算法的分類精度得到瞭提高,特彆是對光譜特徵相似的地物。
고광보도상수거체현위파단다、지물표첨획취곤난대、보신식항간우능력약등특정,용역인기유수재난、광보공간변이성등문제,종이영향분류기적분류정도。침대저사문제,본문장부상사신식인입도랍보랍사지지향량궤(Laplacian Support Vector Machine,LapSVM)적류형정칙화항중,제출료일충인입부상사적랍보랍사지지향량궤(Dissimilarity in Laplacian Support Vector Machine,Diss-LapSVM)분류산법,억제광보공간변이대분류결과적영향。동시,본문제출이용선성근린전파(Linear Neighborhood Propagation,LNP)산법구조도적랍보랍사구진,경유효지인입무표첨양본적신식。실험결과표명,본문산법적분류정도득도료제고,특별시대광보특정상사적지물。
Hyperspectral image has some typical characteristics such as too many of bands,hard to obtaln labeling sam-ples,easy of being interfered of spectral information.These characteristics lead to the dimensionality disaster and the spa-tial variability of spectral information.To solve these problems,it is proposed to introduce dissimilarity in Laplacian support vector machine (Diss-LapSVM)by the adding dissimilarity information to machine’manifold regularization term,which re-stralns the influence of the spatial variability effectively.Meanwhile,in order to introduce appropriately distribution of unla-beled samples,This paper provides linear neighborhood propagation (LNP)to construct graph Laplacian matrix.The re-sults illustrated that the proposed method can improve the classification accuracy,especially for samples which have similar spectral features.