E.A. Samoylin, V.V. Shipko

The method of the interchannel graded-index suppression of pulse interferences in conflict stable systems of processing of color digital images is offered and probed. The purpose of operation is increase of accuracy of restoration of the elements of color digital images distorted by pulse interferences due to property of the interchannel redundance. Now for fight against pulse interferences on color images the vectorial median filters considering the interchannel correlation of values of signals of each component are widely used. Vectorial median filters though show advantage before traditional component-wise processing of the image, at the same time isn't able to recover authentically always value of the lost signal owing to the limitation only selection of elements of a sliding window, and also in case of overwhelming quantity of faulty elements in an aperture. Meanwhile, originating redundance upon transition from gray-scale to color images, opens new opportunities for restoration of the noises of values of signals of images lost owing to influence. In particular, in case of distortion of value of brightness of an image element in one of color channels, it is possible to evaluate this value on amplitude of brightness of appropriate elements of other color channels. The offered method is two-stage. At the 1st stage the spatial provision of noises in each color channel is evaluated. For detection of noises the most various methods and algorithms can be used. At the 2nd stage the interchannel graded-index compensating of the found faulty image elements is carried out. On values not noisy elements in an aperture of the image the interchannel gradients are created (an image element difference in one channel with an element of the adjacent color channel corresponding to coordinate position), without considering the central element. Further estimates of a faulty element of one component on local gradients of other component are created. The assessment is carried out on that component which central element isn't a noise. The general assessment is defined as median line of the local estimates received on all color components. As showed results of the conducted numerical researches, the offered method allows to receive higher accuracy of restoration of the distorted signals in comparison with traditional vectorial median processing.