To use a modified delta algorithm for image processing and coding

Abstract

The paper considers the ideas of using a modified delta algorithm for calculating the differences between adjacent values for image coding as an algorithm for predicting pixel values and for image processing as an algorithm for obtaining image contours. Several modifications and variants of the implementation of these ideas are proposed, which require further research and comparisons in order to determine the most optimal or adaptively optimal for certain conditions, primarily for the use of the delta algorithm when encoding images, since the basic, initial version has an increased dependence and sensitivity to accumulation of errors. Using the block version of the delta algorithm has less error accumulation, but requires more data to be stored. Also considered is a version of the delta algorithm, where the differences are calculated relative to the first value in the block, instead of the sequential difference calculation algorithm. Instead of using only vertical or horizontal differences, their simultaneous and coordinated application is proposed.