METHOD FOR NOISELESS GRADIENT SELECTION OF OBJECT CONTOURS ON DIGITAL HALFTONE IMAGES Russian patent published in 2018 - IPC G06K9/48 G06K9/40 G06K9/56 

FIELD: image processing means.

SUBSTANCE: invention relates to digital image processing and can be used to solve tasks for pattern recognition by their contours in digital images. Disclosed is method for noiseless gradient selection of object contours on digital halftone images. Method comprises a step of determining the luminance gradient modulus at each point of the image and forming contours of the object. Further, method comprises calculating a direct wavelet transform of the rows and columns of the digital halftone image, generating two matrices g_x and g_y of the luminance gradient components at each point of the image by inverse wavelet transformation, wherein analytical functions, which describe the derivatives of the inverse transform wavelets used in the corresponding coordinates, are used as the transformation core.

EFFECT: technical result is higher quality of the generated halftone image contours under interference conditions.

1 cl, 6 dwg