Comparison of Interpolation Methods When Achieving Super-Resolution of Images Based on the Analysis of Several Frames.

Abstract

Increasing the resolution based on the use of multiple frames of the same object leads to additional errors. These errors are the result of errors in determining the position of frames, as well as the result of interpolation inaccuracy. In this paper, only errors caused by interpolation are analyzed. We consider the dependence of errors on the number of frames used, the type of interpolation, the type of the original image. It is shown that after averaging over the positions of individual frames, the key dependencies are the dependence of errors on the number of frames and the type of interpolation. It is shown that it does not make sense to analyze interpolation errors immediately after the interpolation procedure. After the required spectral filtering procedure, relatively small interpolation errors will increase significantly. Change and the type of dependence of errors on the number of frames. Therefore, it is advisable to draw the main conclusions based on the analysis of interpolation errors after the filtering procedure. From the point of view of error minimization, it is preferable to use interpolation by a cubic spline and spectral interpolation.

Restoring Missing Fragments of a Distorted Image Due to Defocusing or Blur of a Known Hardware Function

“…In [14], we have developed a method for interpolation of a sequentially calculated Fourier spectrum (MIPVS), which is capable of filling in gaps of arbitrary shape. In works [15][16][17][18][19], possible areas of application of MIPVS are shown and a comparative analysis of the methods used to fill gaps in images was carried out.…”