METHOD TO AUTOMATICALLY DECODE MICROMATRIX IMAGES Russian patent published in 2012 - IPC G06K9/00 

FIELD: information technologies.

SUBSTANCE: method of automatic identification of corners and probes of a micromatrix chip, even if there are no probes in corners, in a scanned image of a high-density and high-resolution micromatrix, having an image space, in which the method minimises erroneous distortions in the image, includes stages, at which: Radon transform is applied to the input image of the micromatrix for projection of the image into the space of corners and distances, when this is possible, to find orientation of direct lines; fast Fourier transform is applied to the projected image produced at the stage to find the optimum angle of inclination of the projected image; optimal first and last local maxima are detected for the optimal angle of inclination; reverse projection of determined first and last local maximum is carried out into the image space to find, as a first approximation, the first and last vertical lines of the image; the image is rotated; four corners of the image are determined, in a first approximation, from crossing of vertical and horizontal lines; heuristics is applied to determine whether the first approximation is sufficient; and optionally the scanned image is cut around the first approximation of four corners.

EFFECT: expansion of functional capabilities due to application of a multi-stage algorithm of corners finding to the image.

10 cl, 9 dwg, 4 tbl