Identification method and apparatus

Claim 3

A method according to claim 1 wherein the mapping between the plurality of assemblies of arcs extracted from the image of the reference entity to establish the standard level of differences between the said assemblies to establish the criterion of an acceptable level of probability of identity between the two nominally identical entities is done in the same manner as that between the two assemblies of arcs related to linear features of the images of the two nominally identical entities.

Claim 4

A method according to claim 1 wherein the operation of mapping one set of reproduced arcs upon the other is carried out in such a manner as to maximize the correlation between the representations of respective arcs in the assemblies of arcs while simultaneously minimizing elastic distortions in the mapping between the representations of the arcs.

Claim 5

A method according to claim 1, wherein the assembly of arcs is extracted from each image by scanning the image to produce a plurality of pixels, determining whether a pixel chosen at random contains an element of a linear feature of the image and if it does recording the coordinates of the center of that pixel, constructing about the pixel a first area having a size approximately equal to twice the thickness of an expected linear feature of the image and centered upon the pixel, determining a best fit straight line to pixels in the first area containing elements of a linear feature of the image, the said straight line defining a first segment of an extracted arc, estimating therefrom the coordinates of the center of a similar area adjacent the first area and containing another segment of the linear feature of the image represented by the pixels containing elements of the linear feature of the image, determining the best fit straight line to the elements of the linear feature of the image represented by the pixels in the second area to yield a possible second segment of the extracted arc, determining: a) whether the possible second segment of extracted arc is of similar width to the first segment of extracted arc, b) whether the possible second segment of extracted arc joins onto the first segment with a transverse displacement of less than half a line width and c) whether the possible second segment of extracted arc joins the first segment of extracted arc at an angle with reference to the first segment of extracted arc which is within a predetermined value so that the segments of extracted arc at least approximate to an arc of constant curvature; accepting the best fit straight line as representing part of an extracted arc if the above criteria are satisfied, repeating the arc extraction operation in one direction until the above criteria no longer are satisfied, repeating the arc extraction operation in the reverse direction until the above criteria also no longer are satisfied, temporarily storing data generated during the determination of the possible segments of the extracted arc, determining whether the length of the possible extracted arc exceeds a predetermined value and if it does, determining the coordinates of the center of the extracted arc, the angle of the tangent to the extracted arc at its center, the curvature of the extracted arc and the chordal length of the extracted arc, storing these parameters, deleting the pixels from the image, repeating the arc extraction operation in other parts of the image until the entire image has been processed, and storing the above parameters of the extracted arcs to form the said data set relating to the image.

Claim 6

A method according to claim 1 wherein the entities are signatures the first of which is known to be authentic.

Claim 7

A method according to claim 1, wherein the entities to be compared are images of at least portions of faces and there is included the operation of reducing the images to assemblages of curved lines related to localized variations in the shape of regions of the said portions of faces.