METHOD AND DEVICE FOR IDENTIFYING OBJECT IMAGES Russian patent published in 2009 - IPC G06K9/62 

FIELD: physics; computer engineering.

SUBSTANCE: invention relates to computer engineering and can be used in computer vision systems for identifying objects on images. Technical outcome is achieved due to that, the standard is stored not as a series of flat projections, which contain different aspect angles of the object, but its three-dimensional representation in vector form. During identification, the three-dimensional image is rotated, each time with generation of a series of flat images until a match is found. When solving the given task through direct search for all versions, computing time is long, therefore a series of parametres of the object are used, the class of which relates to the given object, overall dimensions. This set of parametres is determined for each model taking into account complexity of its shape. An image is sent to the input of the analyser, presented by an array of pixels in grey gradations, i.e. each element of the array assumes a value from 0 to 255. The array dimension depends on sampling parametres of the image.

EFFECT: increased accuracy of identifying images and provision for invariance to affine transformation of rotation and scaling.

2 cl, 1 dwg