Application of multi-scale singular vector decomposition to vessel classification in overhead satellite imagery

Creation and selection of relevant features for image classification is a process requiring significant involvement of domain knowledge. It is thus desirable to cover at least part of that process with semi-automated techniques capable of discovering and visualizing those geometric characteristics of images that are potentially relevant to the classification objective. In this work, we propose utilizing the multi-scale singular value decomposition (MSVD), which can be efficiently run on large high-dimensional datasets. We apply this technique to create a multi-scale representation of overhead satellite images of various types of vessels, with the objective of identifying those types. We augment the original set of pixel data with features obtained by applying the MSVD to multi-scale patches of the images. The result is then processed using a linear Support Vector Machine (SVM) algorithm. The classification rule obtained is significantly better than the one based on the original pixel space. The generic nature of the MSVD mechanism and standard mechanisms used for classification (SVM) suggest a wider utility of the proposed approach.

[1]  Sing-Tze Bow,et al.  Pattern recognition and image preprocessing , 1992 .

[2]  Peter W. Jones.  Square functions, Cauchy integrals, analytic capacity, and harmonic measure , 1989 .

[3]  Leo Breiman,et al.  Random Forests , 2001, Machine Learning.

[4]  Stéphane Lafon,et al.  Diffusion maps , 2006 .

[5]  David Shallcross,et al.  Centralized multi-scale singular value decomposition for feature construction in LIDAR image classification problems , 2012, 2012 IEEE Applied Imagery Pattern Recognition Workshop (AIPR).

[6]  Milan Sonka,et al.  Image Processing, Analysis and Machine Vision , 1993, Springer US.

[7]  Bernhard Schölkopf,et al.  Nonlinear Component Analysis as a Kernel Eigenvalue Problem , 1998, Neural Computation.

[8]  G. Lerman.  Quantifying curvelike structures of measures by using L2 Jones quantities , 2003 .

[9]  Josh Harguess,et al.  Vessel classification in overhead satellite imagery using learned dictionaries , 2012, Other Conferences.

[10]  M. Maggioni,et al.  Manifold parametrizations by eigenfunctions of the Laplacian and heat kernels , 2008, Proceedings of the National Academy of Sciences.

[11]  Peter W. Jones.  Rectifiable sets and the Traveling Salesman Problem , 1990 .

[12]  V. Rokhlin,et al.  A fast randomized algorithm for overdetermined linear least-squares regression , 2008, Proceedings of the National Academy of Sciences.

[13]  B. Nadler,et al.  Diffusion maps, spectral clustering and reaction coordinates of dynamical systems , 2005, math/0503445.