Unsupervised feature learning for illumination robustness

The illumination conditions of a scene create intra-class variability in outdoor visual data, degrading the performance of high-level algorithms. Using only the image, and with hyper-spectral data as a case study, this paper proposes a deep learning approach to learn illumination invariant features from the data in an unsupervised manner. The proposed approach incorporates a similarity measure, the Spectral Angle, that is relatively insensitive to brightness into the cost function of a Stacked Auto-Encoder so that an illumination invariant mapping is learned from the input data to the hidden layer. Experiments using synthetic and real imagery show that this novel feature learning approach produces a more illumination invariant representation of the data, improving the results of a high-level algorithm (clustering) under such conditions.

[1]  Glenn Healey,et al.  Invariant recognition in hyperspectral images , 1999, Proceedings. 1999 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (Cat. No PR00149).

[2]  Jie Geng,et al.  Hyperspectral image classification via contextual deep learning , 2015, EURASIP Journal on Image and Video Processing.

[3]  Richard F. Gunst,et al.  Applied Regression Analysis , 1999, Technometrics.

[4]  Sildomar T. Monteiro,et al.  Mapping Layers of Clay in a Vertical Geological Surface Using Hyperspectral Imagery: Variability in Parameters of SWIR Absorption Features under Different Conditions of Illumination , 2014, Remote. Sens..

[5]  Xing Chen,et al.  Stacked Denoise Autoencoder Based Feature Extraction and Classification for Hyperspectral Images , 2016, J. Sensors.

[6]  Amin Yazdani Salekdeh.  Multispectral and hyperspectral images invariant to illumination , 2011 .

[7]  Graham D. Finlayson,et al.  Hamiltonian Path based Shadow Removal , 2005, BMVC.

[8]  Cheng Lu,et al.  Intrinsic Images by Entropy Minimization , 2004, ECCV.

[9]  Jon Atli Benediktsson,et al.  Linear Versus Nonlinear PCA for the Classification of Hyperspectral Data Based on the Extended Morphological Profiles , 2012, IEEE Geoscience and Remote Sensing Letters.

[10]  Martin Chamberland,et al.  Chemical agent detection and identification with a hyperspectral imaging infrared sensor , 2007, SPIE Security + Defence.

[11]  Emma Izquierdo-Verdiguier,et al.  Encoding Invariances in Remote Sensing Image Classification With SVM , 2013, IEEE Geoscience and Remote Sensing Letters.

[12]  Carlo Gatta,et al.  Unsupervised deep feature extraction of hyperspectral images , 2014, 2014 6th Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing (WHISPERS).

[13]  Rishi Ramakrishnan,et al.  Shadow compensation for outdoor perception , 2015, 2015 IEEE International Conference on Robotics and Automation (ICRA).

[14]  Sildomar T. Monteiro,et al.  Evaluating Classification Techniques for Mapping Vertical Geology Using Field-Based Hyperspectral Sensors , 2012, IEEE Transactions on Geoscience and Remote Sensing.

[15]  M. Kramer.  Nonlinear principal component analysis using autoassociative neural networks , 1991 .

[16]  Eric O. Postma,et al.  Dimensionality Reduction: A Comparative Review , 2008 .

[17]  Mark S. Drew,et al.  Detecting Illumination in Images , 2007, 2007 IEEE 11th International Conference on Computer Vision.

[18]  Jörgen Ahlberg,et al.  Illumination and shadow compensation of hyperspectral images using a digital surface model and non-linear least squares estimation , 2011, Remote Sensing.

[19]  Fabio Del Frate,et al.  Feature reduction of hyperspectral data using Autoassociative neural networks algorithms , 2009, 2009 IEEE International Geoscience and Remote Sensing Symposium.

[20]  Wilhelm Burger,et al.  Digital Image Processing - An Algorithmic Introduction using Java , 2008, Texts in Computer Science.

[21]  Amit Banerjee,et al.  Hyperspectral video for illumination-invariant tracking , 2009, 2009 First Workshop on Hyperspectral Image and Signal Processing: Evolution in Remote Sensing.

[22]  S. J. Sutley,et al.  USGS Digital Spectral Library splib06a , 2007 .

[23]  Gang Wang,et al.  Deep Learning-Based Classification of Hyperspectral Data , 2014, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.