High Resolution Bathymetry Estimation Improvement with Single Image Super-Resolution Algorithm "Super-Resolution Forests"

Abstract

Using the single image super-resolution algorithm Super-Resolution Forests (SRF), this paper shows the ability to improve the prediction of high resolution bathymetry data. Borrowing the machine-learning technique of training and testing on a dictionary of sets data, we could create high resolution estimates of bathymetry data similar to estimates typically created with this technique using image data. By implementing a changed variance on the training process of the SRF algorithm, we were able to further increase the mean PSNR score of the high resolution estimated data from previously used bicubic interpolation.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 26, 2017
Accession Number
AD1028639

Entities

People

  • David Bonanno
  • Dylan Einsidleg
  • Frederick E. Petry
  • Kristen Nock
  • Leslie Smith
  • Paul A Elmore
  • Warren T. Wood

Organizations

  • United States Naval Research Laboratory

Tags

Communities of Interest

  • Materials and Manufacturing Processes
  • Space

DTIC Thesaurus Topics

  • Abstracts
  • Algorithms
  • Artificial Intelligence
  • Bathymetry
  • Department Of Defense
  • Frequency
  • High Resolution
  • Information Operations
  • Information Systems
  • Interpolation
  • Low Resolution
  • Machine Learning
  • Mathematics
  • Military Research
  • Seabed
  • Training

Fields of Study

  • Computer science

Readers

  • Computer Vision.
  • Nanoscale Plasmonic Nanotechnology
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Neural Networks