Hyperspectral data processing techniques for geological applications


cr

Space-borne hyperspectral remote sensing data find use in several domains such as land use - land cover mapping, agricultural studies, environmental and atmospheric studies and geological studies. However high spectral resolutions imply difficulties in maintaining sufficient electromagnetic energy for good signal-to- noise ratio at narrow band widths, which result in significant noise in hyperspectral datasets. Hyperspectral datasets are also affected by spectral noise such as smile effect. Very often good signal-to- noise ratios are obtained at the expense of spatial resolution, which leads to mixed spectral response from the pixels and errors in classification outputs.


This research aims at developing new algorithms for reducing spectral and spatial noise in hyperspectral data, as well as new spectral unmixing techniques particularly for geological mapping. A secondary goal is to suppress vegetation signal in the pixel response.




Key Research Interests

Development of hyperspectral data processing techniques for mineral and lithological mapping, which includes, pre-processing techniques for noise reduction (random/salt-and- pepper noise and striping), illumination effects correction, topographic correction, atmospheric correction, spectral noise reduction and data size and dimensionality reduction, and post-processing techniques for vegetation suppression and unveiling surface lithology, and image classification for mineral and lithologic mapping.


The key outcome of the research would be identification and quantification of targeted surficial lithology and minerals, which can assist in narrow down the targeted area for further study in mineral exploration.




Publications

Journals


  • Pal, M. K., & Porwal, A. (2015). A Local Brightness Normalization (LBN) algorithm for destriping Hyperion images. International Journal of Remote Sensing, 36(10), 2674-2696.
  • Pal, M. K., Porwal, A. (2016). Across-Track- Illumination Correction using Quadratic-Least-Square-Regression from Hyperion image, Photogrammetry and Remote Sensing (under review).

Conferences


  • Pal M. K., Porwal A., 2015. Spectral noise reduction and smoothing using local cubic least square regression from Hyperion reflectance data, 2015 IEEE 2nd International Conference on Signal Processing and Integrated Networks, pp. 752-755, 19-20 February, 2015, Amity University, Noida, Uttar Pradesh, India.
  • Pal M. K., Porwal A., 2015. Destriping of Hyperion Images Using Low-Pass- Filter and Local-Brightness- Normalization, 2015 IEEE International Geoscience and Remote Sensing Symposium, 26 July - 31 July 2015, Milan, Italy.
  • Mahendra K. Pal, Alok Porwal. Dimensionality Reduction of Hyperspectral Data: Band Selection Using Curve Fitting, SPIE Asia-Pacific Remote Sensing, 4 - 7 April 2016, New Delhi, India.
  • Pal M. K., Porwal A., 2016. Vegetation Suppression for Unveiling of Surface Lithology from Hyperspectral Images Using Linear Spectral Unmixing Approach, 2016 IEEE International Geoscience and Remote Sensing Symposium, 10 July - 15 July 2016, Beijing, China (accepted).
  • Pal M. K., Porwal A., 2016. Optimizing Classification Using Multi-Classifiers for Spaceborne Hyperspectral Dataset, WHISPERS-2016, 21 st -24 th August 2016, Los Angeles,California, USA (accepted).