A comparative analysis of classification methods for hyperspectral images generated with conventional dimension reduction methods

This paper compared performances of classification methods for a hyperspectral image dataset in view of dimensionality reduction (DR). Among conventional DR methods, principal component analysis, maximum noise fraction, and independent component analysis were used for the purpose of dimension reduction. The study was conducted using these DR techniques on a real hyperspectral image, an AVIRIS dataset with 224 bands, throughout the experiments. It was observed that DR may have a significant effect on the classification performance. After the DR methods were applied to the image dataset, the extracted reduced bands were used for testing classification performances. Four commonly used classification methods including maximum likelihood, Mahalanobis distance, spectral angle mapper, and support vector machines were used for the classification of dimensionally reduced and original (not reduced) images to test whether classification accuracies differed significantly for these images. Experiments reported in this study indicate that second-order statistics-based DR methods could efficiently reduce the dimensionality of the hyperspectral dataset. The maximum likelihood method generally performed better than the other classifiers while the spectral angle mapper exhibited significantly lower accuracies in the experiments. Inclusion of a texture measure in classification improved the classification results in a significant manner.

PDF

___

[1] Plaza A, Martinez P, Plaza J, Perez RM. Dimensionality reduction and classification of hyperspectral image data using sequences of extended morphological transformations. IEEE T Geosci Remote 2005; 43: 466-479.
[2] Junying S, Ning S. A dimensionality reduction algorithm of hyper spectral image based on fract analysis. In: 21st ISPRS Congress Technical Commission VII, International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences; 311 July 2008; Beijing, China. pp. 297-302.
[3] Hsu PH, Tseng YH, Gong P. Dimension reduction of hyperspectral images for classification applications. Annals of GIS 2002; 8: 1-8.
[4] Fong M. Dimension Reduction on Hyperspectral Images. Los Angeles, CA, USA: UCLA Department of Mathematics, 2007.
[5] Green AA, Berman M, Switzer P, Craig MD. A transformation for ordering multispectral data in terms of image quality with implications for noise removal. IEEE T Geosci Remote 1988; 26: 65-74.
[6] Gordon C. A generalization of the maximum noise fraction transform. IEEE T Geosci Remote 2000; 38: 608-610.
[7] Roger RE. Principal components transform with simple, automatic noise adjustment. Int J Remote Sens 1996; 17: 2719-2727.
[8] Greco M, Diani M, Corsini G. Analysis of the classification accuracy of a new MNF based feature extraction algorithm. In: SPIE 2006 Image and Signal Processing for Remote Sensing XII Conference; 11 September 2006; Stockholm, Sweden.
[9] Liu X, Zhang B, Gao L, Chen D. A maximum noise fraction transform with improved noise estimation for hyperspectral data. Sci Chin Ser F-Inf Sci 2009; 52: 1578-1587.
[10] Bayliss J, Gualtieri JA, Cromp RF. Analyzing hyperspectral data with independent component analysis, In: SPIE 1997 26th AIPR Workshop: Exploiting New Image Sources and Sensors Conference; 15 October 1997; Washington DC, USA. pp. 133-143.
[11] Wang J, Chang C. Independent component analysis-based dimensionality reduction with applications in hyperspectral image analysis. IEEE T Geosci Remote 2006; 44: 1586-1600.
[12] Richards JA. Remote Sensing Digital Image Analysis. Berlin, Germany: Springer-Verlag, 1999.
[13] Kruse FA, Lefkoff AB, Boardman JB, Heidebrecht KB, Shapiro AT, Barloon PJ, Goetz AFH. The spectral image processing system (SIPS) - interactive visualization and analysis of imaging spectrometer data. Remote Sens Environ 1993; 44: 145-163.
[14] Hsu CW, Chang CC, Lin CJ. A Practical Guide to Support Vector Classification. Taipei, Taiwan: National Taiwan University, 2007.
[15] Aky¨urek O. A comparative analysis of dimension reduction methods in hyperspectral images . MSc, Kocaeli ¨ University, Kocaeli, Turkey, 2012 (in Turkish).
[16] Cattell RB. The scree test for number of factors. Multivar Behav Res 1966; 1: 245-276.
[17] Congalton RG, Green K. Assessing the Accuracy of Remotely Sensed Data: Principles and Practices. Boca Raton, FL, USA: Lewis Publishers, 1999.
[18] Keshava N, Kerekes J, Manolakis D, Shaw G. An algorithm taxonomy for hyperspectral unmixing. In: Proceedings of the SPIE AeroSense Conference on Algorithms for Multispectral and Hyperspectral Imagery VI; 2000. pp. 42-63.
[19] Nascimento J, Bioucas-Dias JM. Does independent component analysis play a role in unmixing hyperspectral data? IEEE T Geosci Remote 2005; 43: 175-187.