Change detection in Arctic satellite imagery using clustering of sparse approximations (CoSA) over learned feature dictionaries

Advanced pattern recognition and computer vision algorithms are of great interest for landscape characterization, change detection, and change monitoring in satellite imagery, in support of global climate change science and modeling. We present results from an ongoing effort to extend neuroscience-inspired models for feature extraction to the environmental sciences, and we demonstrate our work using Worldview-2 multispectral satellite imagery. We use a Hebbian learning rule to derive multispectral, multiresolution dictionaries directly from regional satellite normalized band difference index data. These feature dictionaries are used to build sparse scene representations, from which we automatically generate land cover labels via our CoSA algorithm: Clustering of Sparse Approximations. These data adaptive feature dictionaries use joint spectral and spatial textural characteristics to help separate geologic, vegetative, and hydrologic features. Land cover labels are estimated in example Worldview-2 satellite images of Barrow, Alaska, taken at two different times, and are used to detect and discuss seasonal surface changes. Our results suggest that an approach that learns from both spectral and spatial features is promising for practical pattern recognition problems in high resolution satellite imagery.

[1]  C. Symon,et al.  Arctic climate impact assessment , 2005 .

[2]  Neal R. Harvey,et al.  Comparison of GENIE and conventional supervised classifiers for multispectral image feature extraction , 2002, IEEE Trans. Geosci. Remote. Sens..

[3]  Daniela I. Moody,et al.  Change detection and classification of land cover in multispectral satellite imagery using clustering of sparse approximations (CoSA) over learned feature dictionaries , 2014, 2014 IEEE Applied Imagery Pattern Recognition Workshop (AIPR).

[4]  Daniela I. Moody,et al.  Undercomplete learned dictionaries for land cover classification in multispectral imagery of Arctic landscapes using CoSA: clustering of sparse approximations , 2013, Defense, Security, and Sensing.

[5]  Daniela I. Moody,et al.  Land cover classification in multispectral imagery using clustering of sparse approximations over learned feature dictionaries , 2014 .

[6]  Daniela I. Moody,et al.  Learning sparse discriminative representations for land cover classification in the Arctic , 2012, Optics & Photonics - Optical Engineering + Applications.

[7]  Mike E. Davies,et al.  Gradient Pursuits , 2008, IEEE Transactions on Signal Processing.

[8]  A. Lewkowicz,et al.  Frequency and magnitude of active‐layer detachment failures in discontinuous and continuous permafrost, northern Canada , 2005 .

[9]  Rasim Latifovic,et al.  Recent (1986 - 2006) Vegetation-Specific NDVI Trends in Northern Canada from Satellite Data , 2009 .

[10]  Guillermo Sapiro,et al.  Discriminative learned dictionaries for local image analysis , 2008, 2008 IEEE Conference on Computer Vision and Pattern Recognition.

[11]  Daniela I. Moody,et al.  Unsupervised land cover classification in multispectral imagery with sparse representations on learned dictionaries , 2012, 2012 IEEE Applied Imagery Pattern Recognition Workshop (AIPR).

[12]  Kenneth M. Hinkel,et al.  Spatial Extent, Age, and Carbon Stocks in Drained Thaw Lake Basins on the Barrow Peninsula, Alaska , 2003 .

[13]  Bruno A. Olshausen,et al.  Learning sparse, overcomplete representations of time-varying natural images , 2003, Proceedings 2003 International Conference on Image Processing (Cat. No.03CH37429).

[14]  R. Myneni,et al.  The interpretation of spectral vegetation indexes , 1995 .

[15]  Claudio Parente,et al.  Coastline extraction using high resolution WorldView-2 satellite imagery , 2014 .

[16]  Brendt Wohlberg,et al.  Inpainting by Joint Optimization of Linear Combinations of Exemplars , 2011, IEEE Signal Processing Letters.

[17]  Kjersti Engan,et al.  Image compression using learned dictionaries by RLS-DLA and compared with K-SVD , 2011, 2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[18]  Susan S. Hubbard,et al.  Quantifying and relating land-surface and subsurface variability in permafrost environments using LiDAR and surface geophysical datasets , 2012, Hydrogeology Journal.

[19]  Douglas J. King,et al.  Estimating the extent of near‐surface permafrost using remote sensing, Mackenzie Delta, Northwest Territories , 2009 .

[20]  Kjersti Engan,et al.  Method of optimal directions for frame design , 1999, 1999 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings. ICASSP99 (Cat. No.99CH36258).

[21]  Andrew Y. Ng,et al.  The Importance of Encoding Versus Training with Sparse Coding and Vector Quantization , 2011, ICML.

[22]  Sarah E. Gergel,et al.  Spatial Heterogeneity in the Shrub Tundra Ecotone in the Mackenzie Delta Region, Northwest Territories: Implications for Arctic Environmental Change , 2010, Ecosystems.

[23]  F. Stuart Chapin,et al.  Detecting changes in arctic tundra plant communities in response to warming over decadal time scales , 2004 .

[24]  Antonio F. Wolf,et al.  Using WorldView-2 Vis-NIR multispectral imagery to support land mapping and feature extraction using normalized difference index ratios , 2012, Defense + Commercial Sensing.

[25]  M. Elad,et al.  $rm K$-SVD: An Algorithm for Designing Overcomplete Dictionaries for Sparse Representation , 2006, IEEE Transactions on Signal Processing.

[26]  Peter Kuhry,et al.  High‐resolution mapping of ecosystem carbon storage and potential effects of permafrost thaw in periglacial terrain, European Russian Arctic , 2011 .

[27]  Garrett T. Kenyon,et al.  Large-scale functional models of visual cortex for remote sensing , 2009, 2009 IEEE Applied Imagery Pattern Recognition Workshop (AIPR 2009).

[28]  Vincent Lepetit,et al.  Are sparse representations really relevant for image classification? , 2011, CVPR 2011.

[29]  Daniela I. Moody,et al.  Land cover classification in multispectral satellite imagery using sparse approximations on learned dictionaries , 2014, Sensing Technologies + Applications.

[30]  A. Bruckstein,et al.  K-SVD : An Algorithm for Designing of Overcomplete Dictionaries for Sparse Representation , 2005 .