Noise suppression in multichannel video sequences using fuzzy logic theory and local adaptive filtering

A study was conducted to propose a filtering algorithms taking into account the inter-channel and inter-frame correlations in the video sequence (VS) based on a hybrid approach, which combined the methods of fuzzy logic theory and the adaptive filtering on the basis of estimates of local statistical properties in the color-VS frames. These algorithms implemented the processing of images on the basis of pixel gradients in different directions when filtering jointly the VS neighboring frames while using the new fuzzy-logic rules singling out pixels with similar structural properties and increasing the sampling volume and improving the filtering quality significantly. The criteria used when comparing the proposed algorithms with those known from the literature were the peak signal-to-noise ratio (PSNR) in decibels, the value of the mean absolute error (MAE) determining the quality of reconstruction of fine details, the normalized color difference (NCD) describing the chromatic properties, and the estimate of the structural similarity index measure (SSIM) index.