METHOD OF IDENTIFYING OBJECTS ON DIGITAL IMAGES OF AN UNDERLYING SURFACE BY FUZZY TRIANGULATION OF DELAUNAY Russian patent published in 2020 - IPC G06K9/00 

FIELD: computer equipment.

SUBSTANCE: invention relates to computer science and computer engineering. Technical result is achieved by the fact that preliminary processing of the initial image is performed, bringing the image of the associated set of objects introduced into the computing device to the normal scale, which is standard for the given modification method, alignment, insertion of the required size into the rectangle, successive comparison with stored in the computer memory templates, which are stored in form of fuzzy Delaunay triangulations, which are compared to a fuzzy Delaunay triangulation supplied to the scene image input by means of the neural network. Comparison is carried out by analyzing features of the form of triangulation of the obtained scene of the Earth's surface objects, wherein comparison is made for each feature, and a decision is made on coincidence of the Delaunay triangulation of the vector model of the scene of the obtained image and reference fuzzy triangulations of Delaunay.

EFFECT: high reliability and accuracy of identifying a spatial scene in a robot vision system and determining the location and position of the target object in a local coordinate system.

4 cl, 3 dwg