Constructing FIR-filters in a given parametrical class of frequency response for defocus correction

A problem of the restoration of defocused images is considered. For this purpose, we construct an FIR-filter by identifying the model from a special class of parameters with use of test images. The model of impulse response is defined in a class of univariate functions, which approximate the required frequency response in the radial direction. Samples of the two-dimensional impulse response are defined by discretization and the subsequent sample normalization. The approximation function class is defined so as to amplify low and suppress high spectrum frequencies. Important advantages of the method are high-quality image restoration and fast identification of filter's model due to the fact that the approximating function is determined by a small number of unknown parameters. In this article realization examples are given. These examples show the possibility of achieving the higher-quality restoration in comparison with the Wiener filter from the open-source image processing library OpenCV.