Algorithm for recognizing geometric objects in video images using the Haar Cascade method

The article discusses linear image filtering methods based on the solution of the Wiener–Hopf equation, used to suppress noise and restore degraded images. Analytical expressions of filters minimizing the root-mean-square error of reconstruction are presented, and their effectiveness is compared with classical Wiener filters. An algorithm for constructing an optimal linear filter in the spatial and frequency domains has been developed, taking into account the correlation properties of the signal and noise. Modeling using synthetic and real images has been carried out, confirming the high accuracy of reconstruction and the stability of the proposed method under various types of noise distortion.