Adaptive integer arithmetic coding with RLE-transform

Abstract

The paper proposes a method of multilevel selective processing to increase the availability of information about critical infrastructure objects with a given level of reliability and confidentiality. This technology is based on the identification of key information at several stages of processing and adaptation of the RLE algorithm and integral arithmetic coding into a new input data structure. Thus, the last stage of processing video data by the proposed approach is considered, which is aimed at reducing the volume with the following features: further development of the method of linearization of two-dimensional transforms based on zig-zag scanning is obtained. The differences of the method lie in carrying out the vector intertransform zigzag linearization taking into account the selection of spectral components, which is defined as complementary. For the first time, a method of decomposition of a linearized transform based on the definition of a threshold has been developed. Characteristic features of the method: the threshold is found taking into account the presence in the group of different types of transformants to determine the total uneven number of non-equilibrium complementary components; the integral arithmetic coding has been improved based on taking into account the frequencies of dictionary elements (two-dictionary integral arithmetic coding). Distinctive features of the method: determination of the current code components by the decomposed working interval, depending on the capacity of the dictionaries of significant elements and the number of repetitions. This allows additionally taking into account the statistical features of the components of the RLE-structured linearized transform and reducing the length of the arithmetic code; For the first time, a transformant compression technology was created based on the reduction of various types of redundancy in groups of transformants. Characteristic features of the technology: RLE-structuring of the non-uniform complementary part of the group of transforms after their vector linearization and decomposing; the formation of two arithmetic codes according to two dictionary principles of uneven division of the working interval, depending on the type of transformants in the group.