The development of the computational method for determining the constants of recalculation of digital text images on the core image

The article proposes a computational method for determining the constants of recalculation of digital text images on the core image. There is a contradiction between the need to use a cube model of constant of recalculation and a small number of reference pairs "symbol of the current frame - a symbol of the core frame" in the early stages of characters identifying of current and base frames. To avoid this contradiction the article offers multistage increase in the order of constants recalculation (with specification of their parameters) and a simultaneous increase in the amount used for the calculation of reference symbol pairs. To ensure substantially the same accuracy of estimates of coordinates character on the entire frame reference symbols are evenly distributed on the frame, and from the plurality of the candidates excluded assessment of provisions similar characters. Reference symbols are selected as the pair "a symbol of the current frame - a symbol of the core frame" module of total deviation between the estimates of the coordinates that are less than the corresponding mean values in the selected set of identified pairs. Also, the article provides accuracy determination parameters of character coordinates on the core frame as reference characters, and all other symbols on the digital image.