Chapter 1: Invited Addresses and Tutorials on Signals, Coding,
  Systems and Intelligent Techniques

Title: High resolution image restoration by new super-resolution algorithm with context switching motion estimator and improved POCS based reconstruction stage

Author(s): Krzysztof Malczewski, Ryszard Stasinski

Address: Institute of Electronics and Telecommunications, Poznan University of Technology Piotrowo, 3A, 60-965 Poznan, Poland | Institute of Electronics and Telecommunications, Poznan University of Technology Piotrowo, 3A, 60-965 Poznan, Poland

Reference: 12th International Workshop on Systems, Signals and Image Processing pp. 407 - 410

Abstract/Summary: In the paper Super-Resolution algorithm with enhanced reconstruction stage and the context switched motion estimation is presented. The switching motion estimation concept stems from the simple premise that Fourier analysis method may be not accurate when the spectrum is degraded by aliasing. In such situations Taylor series expansion approximations is favourable. In the frequency domain based motion estimation the properties of relocated images Fourier transforms are used to estimation of rotation and translation, as well as sub-sampling artefacts. In this paper we propose a concept based on switching motion estimator between spectrum and Taylor series methods. The choice of one of them depends on an aliasing factor, which we can evaluate. Previously, in the image reconstruction phase the bicubic interpolation has been usually employed, in this paper it is replaced by iterative projection onto convex sets (POCS) with biharmonic spline interpolation approach at its core. The new algorithm indeed gives significant improvement of results if compared to the previous ones.

Order a copy of this article