Спосіб розпізнавання плоских геометричних фігур

Abstract

The proposed method for identifying plane geometric figures involves scanning the image of an original figure, converting the figure image into a series of electric signals, determining the area of the common intersecting part of the original figure and its copy, performing parallel step-by-step translation of the figure copy, and determining, at each translation step, the area of the common intersecting part of the original figure and the copy until this area will be zero. This state corresponds to the maximal shift of the figure copy in the first direction of the translation. For this direction, the function characterizing the intersection of the original figure and its copy is determined. Then, in the same manner, the figure copy is translated in the second direction that is perpendicular to the first direction and the function characterizing the intersection of the original figure and its copy is determined for the second direction. The characteristic function values are compared with specified reference values, and using the results of the comparison, classes X and Y of the figures are determined for the corresponding directions of the translation. An unknown figure is identified by the intersection of the figure sets belonging to the X and Y classes. The proposed method allows a number of translation directions to be reduced.