基于分類別PCA散度的高光譜圖像分類波段選擇
Band Selection Using Divergence of Class-within PCA in Hyperspectral Images Classification
-
摘要: 波段選擇是去除高光譜圖象段間冗余，實(shí)現(xiàn)降維的有效方法。該文提出了一種新的基于分類別主成分分析(PCA)散度的波段選擇方法。即首先對訓(xùn)練集各類樣本分別進(jìn)行PCA變換去相關(guān)并計(jì)算散度，接著分析相應(yīng)PCA變換系數(shù)獲得對各類樣本分類都重要的原始波段，在綜合考慮波段的相關(guān)度，散度和子集規(guī)模的基礎(chǔ)上獲得最終選擇波段。復(fù)雜度分析表明該方法較局部尋優(yōu)的前向搜索計(jì)算量大為降低，提高了效率，并用高光譜遙感圖象的分類實(shí)驗(yàn)進(jìn)行了驗(yàn)證。
-
關(guān)鍵詞:
- 高光譜圖像分類;波段選擇;分類別PCA;散度
Abstract: Band selection from multispectral or hyperspectral image data is an effective method to remove redundancy among bands and thus reduce dimension. An efficient algorithm using divergence based class-within principal component analysis (PCA) and analysis of corresponding coefficients is proposed. At first, the covariance of each class is diagonalized through PCA transforms on class data respectively, and then the divergence only depends on the summation of individual feature separability of transformed bands. Secondly, after an analysis of corresponding PCA transform coefficients, the candidate bands, original bands essential to classification, are determined by majority vote. At last, the final band subset is obtained by analyzing the dependency and divergence of bands in every subset generated according to the correlations of original band in candidates. Compared with sequential forward selection, the proposed method reduces the computation complexity, and encouraging results have been shown by experiments with an Airborne Visible/InfraRed Imaging Spectrometer (AVIRIS) data set. -
Velez-Reyes M, Linares D M. Comparison of principal-compon-[2]ent-based band selection methods for hyperspectral imagery. Image and Signal Processing for Remote Sensing VII, Proc[J].SPIE.2002, 4541:361-369[3]Withagen Paul J, Breejen Eric den, et al.. Band selection from a hyperspectral data-cube for a real-time multispectral 3CCD camera. Algorithms for Multispectral, Hyperspectral, and Ultraspectral Imagery VII, Proc. SPIE AeroSense, 2001, 4381: 84 93.Sheffer D, Ultchin Y. Comparison of band selection results using.[4]different class separation measures in various day and night conditions. Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery IX, Proc[J].SPIE.2003, 5093:452-461[5]Swain P H, King R C. Two effective feature selection criteria for multispectral remote sensing. First International Joint Conference on Pattern Recognition, Washington, DC, 1973: 536-540.[6]J. P. Marques de s著，吳逸飛譯. 模式識別原理、方法及[7]應(yīng)用. 北京: 清華大學(xué)出版社, 2002 : 116-118.[8]Chang Chein-I, Du Qian, et al.. A joint band prioritization and band-decorrelation approach to band selection for hyperspectral[9]image classification. IEEE Trans[J].on Geoscience and Remote Sensing.1999, 37(6):2631-2641 -
計(jì)量
- 文章訪問數(shù): 2801
- HTML全文瀏覽量: 125
- PDF下載量: 1122
- 被引次數(shù): 0
下載:
