Estimation of super-resolution imaging errors based on the use of multiple frames

This work shows the possibility of estimating errors of super-resolution imaging based on the use of several frames. The possibility of identifying areas with large interpolation errors is shown, based only on the interpolation method and interpolated data used. The possibility of determining the necessary number of frames used for imaging from the interpolation error estimate is demonstrated. In this paper, we present both the results of interpolation of the generated test frames, the method of correlation subpixel detection of frame shifts relative to 1 frame, and the estimates of the resulting interpolation errors. It is shown that before measurements, the number of frames necessary to achieve the required level of interpolation errors can not be predicted. The proposed method allows the error value to be controlled during the measurement. It is shown that the estimation of errors of the super resolution method based on test images is not very promising, since the error value depends both on the image itself and on a set of subpixel offsets of low-resolution frames. We discuss the entire procedure of obtaining the results of the work, starting from the test image generation and ending with the conclusions.