302 
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008 
As table 3 shown, the correlation of spectral curve among the 
centre pixel and each pixel in texture unit has obvious 
difference especially for the resident texture unit. And different 
object has very similar correlation which will lead the pixel 
confuse for the pixel classification. The fractal dimension value 
can obtain better result for the classification. The different 
object has different fractal dimension value of spectral curve in 
texture unit. The resident object has the fractal dimension value 
ranged with 1.0199 ~ 1.0271, the tree ranged with 1.0152 ~ 
1.0210 and water ranged with 1.0105 ~ 1.076. Thus the 
dimensionality reduction result using the fractal feature of 
spectral curve can realize better texture code matching which is 
very useful for image classification. 
For the Huges phenomenon, there are still some of different 
object fractal dimension feature while the fractal dimension of 
the centre pixel is quite difference. The dimensionality 
reduction based on spectral curve fractal analysis can combine 
the spectral information and spatial texture information together 
to realize the feature analysis and it can obtain better processing 
efficiency of hyper spectral data. The dimensionality reduction 
based on spectral curve fractal analysis provide a new method 
to differ the confuse pixel such as the feature analysis combined 
the spatial fractal analysis and spectral fractal. 
ACKNOWLEDGEMENTS 
The research work is supported by the Nation Science Founder 
of China. (NSFC: 40371079) and 973 Project of China 
(2006CB701303). 
REFERENCES 
C. Lee and D. A. Landgrebe, 1993. Analyzing high-dimensional 
multi-spectral data, IEEE Trans. Geosci. Remote Sensing, vol. 
31, pp. 792-800. 
A. Jain and D. Zongker, 1997. Feature selection: evaluation, 
application, and small sample performance, IEEE Trans. Pattern 
Anal. Machine Intell.,vol. 19, pp. 153-158. 
L. Bruzzone, F. Roli, and S. B. Serpico,1995. An extension to 
multiclass cases of the Jeffreys-Matusita distance, IEEE Trans. 
Geosci. Remote Sensing, vol. 33, pp. 1318-1321. 
L. Bruzzone and S. B. Serpico, 2000. A technique for feature 
selection in multiclass cases, Int. J. Remote Sensing, vol. 21, pp. 
549-563. 
C. Lee and D. A. 1993. Landgrebe Analyzing high-dimensional 
multispectral data, IEEE Trans. Geosci. Remote Sens., vol. 31, 
pp. 792-803 
CM Chang and S. Wang, 2006. Constrained band selection for 
hyperspectral imagery, IEEE Trans. Geosci. Remote Sens., vol. 
44, pp. 1575-1585. 
A. Plaza,P. Martinez, J. Plaza and R.2005. Perez 
Dimensionality reduction and classification of hyper spectral 
image data using sequences of extended morphological 
transformations, IEEE Trans. Geosci. Remote Sens., vol. 43, pp. 
466-478. 
Qiu H L, Lam N S N, Quattrochi D A, 1991. Fractal 
characterization of hyper spectral imagery[J], Photogrammetric 
Engineering and Remote Sensing, 65: 63~71 
Weng Q., 2003. Fractal analysis of satellite-detected urban heat 
island effect [J], Photogrammetric Engineering and Remote 
Sensing, 69: 555~566 
Peter F, Tommy L, Catherine O., 2001. Statistical analysis of 
hyper spectral data from two Swedish lakes[J].The Science of 
the Total Environment,268:155~169. 
Shu N., 2001.Edge Extraction from Multi-spectral Images and 
Density Analysis of Super-dimensional Spectral Space. 
Proceedings of SPIE,4 550:63~66 
Liu Juan and Moulin P. 2001. Information-theoretic analysis of 
inter-scale and intra-scale dependencies between image wavelet 
coefficients [Jj.IEEE Transaction on Image Proc., 
10(11):1647~1658. 
Berkner K and Wells R O.,2002. Smoothness estimates for soft- 
threshold denoising via translation invariant wavelet transforms 
[J].Applied and Computational Harmonic Analysis, 12(1):1~ 
24. 
Jansen M, 2001. Noise reduction by wavelet threshold 
[M],Springer Verlag, Lecture notes in Statistics (161). 
Wu C. Q., 2004. Adaptive filtering in spatial domain based on 
spectral information [J]. J. Remote Sensing, 2004.8(1):51~55 
Pan Quan, Dai Guanzhong, Zhang Hongcai,1998. A Threshold 
selection method for hard-threshold Filtering Algorithm [J], 
Acta Electronic Science„26(l):l 15'— 117. 
Pan Quan, Meng J.L., Zhang L., Cheng Y.M., Zhang H.C.,2007. 
Wavelet Filtering Method and Its Applications [J], J. Electronic 
and Information Technology, 29(1): 236~242