Method of compression and ensuring the fidelity of video images in infocommunication networks

Subject research in the article is the methods of compressing video images under conditions of ensuring the desired level of their fidelity in the delivery process using infocommunication networks. The goal is to develop methods of encoding video images for increasing the level of their compression in the conditions of ensuring required reliability. Task: to substantiate the approach regarding the structural clusterization of transformed video segments in the conditions of preserving their reliability; to develop a method of structural and statistical coding of transformants in the spectral-cluster space; conduct a comparative evaluation of the effectiveness of various methods of encoding video segments. The methods used: mathematical models for estimating the amount of statistical and structural redundancy in the clustered spectral space of video segments; methods of statistical coding. The following results have been obtained. The potential effectiveness of representing a transformant in clustered space by the number of series of units in binary description of their components has been substantiated. A method of structural-statistical coding in the spectral-cluster space has been created. The basic component of this technological approach is the evaluation of the estimates regarding the potential ability to eliminate various types of redundancy in the current cluster. Here, the amount of redundancy is reduced considering the statistical and structural features of the cluster. The comparative evaluation revealed the advantages of the created method over coding methods in standardized platforms. The advantage is achieved in terms of the peak signal-to-noise ratio by at least 30%. and in terms of a compression ratio by an average of 12 %. Conclusions. The scientific novelty of the obtained results is as follows: for the first time, a method of structural-statistical coding of video segments in spectral space based on their clusterization has been created. The differences of the method lie in the fact that the component of the transformant is simultaneously interpreted as an element of the statistical and combinatorial cluster space; the potential capabilities of eliminating various types of redundancy in the clustered transformant are considered. This provides an increase in the level of compression of video images for a given level of reliability.