Digital Deconvolution: Image Sampling and Restoration Techniques.

Abstract

The finite length digital deconvolution problem is formulated and discussed in terms of modern optimization theory. The ill-conditioned nature of deconvolution is identified and classic deconvolution operators are examined in terms of their respective effectiveness and ease of implementation. Continuous deconvolution filters that are optimal in terms of minimum variance and solution smoothness are derived. these filters contain digital counterparts which can be obtained by means of conventional frequency sampling techniques. Quantization and frequency-aliasing are identified as characteristic noise sources of digital deconvolution and analyzed in terms of their respective corrupting effects upon accurate optical signal estimation. Three approaches for eliminating frequency-aliasing error are discussed. A multiple intensity measurement technique for determining the magnitude and phase of complex optical wavefronts by means of a coherent optics-microprocessor hybrid system is developed. A picture-shift sampling scheme is proposed for effectively increasing picture sampling resolution without requiring additional computer core. A method is developed, based upon this scheme and incoherent light illumination, for performing continuous lowpass filtering.

Document Details

Document Type
Technical Report
Publication Date
Oct 01, 1975
Accession Number
ADA022455

Entities

People

  • Daniel J. Udovic
  • Raj Mittra

Organizations

  • University of Illinois Urbana–Champaign

Tags

DTIC Thesaurus Topics

  • Computers
  • Computing Devices
  • Filters
  • Filtration
  • Frequency
  • Hybrid Systems
  • Illumination
  • Intensity
  • Measurement
  • Microprocessors
  • Optics
  • Optimization
  • Sampling
  • Wavefronts

Fields of Study

  • Engineering
  • Physics

Readers

  • Image Processing and Computer Vision.
  • Operations Research
  • Regression Analysis.