Claims Analysis
Total Claims: 9
Claim 1
Independent Claim
417 words
A method for determining at least one reference autograph character string as a reference set based on a sample set of sample autograph character strings from a same writer, comprising the steps of: through a writing by hand of a predetermined text, producing at least three sample character strings by a same writer for formation of the sample set; during the production of the sample character strings using a stylus, recording and retaining characteristic dynamic parameters describing motion of the stylus; on the basis of the dynamic parameters, and in dependence on whether the stylus is pressed against a writing surface or not, decomposing each sample character string into at least a first and a second partial character string in the form of a first and a second stroke; normalizing the dynamic parameters of all the character strings; determining and retaining differences in the dynamic parameters between a respective sample character string and all sample character strings from the sample set not as yet taken into account in the determination for each dynamic parameter as a parameter deviation; carrying out the determining and retaining of the differences in the dynamic parameters strokewise such that strokes are compared via dynamic programming; determining that sample character string as said at least one reference character string for the reference set which comprises a smallest parameter deviation from all other sample character strings with respect to all dynamic parameters; during calculation of minimal distance between respective strokes, a combination of at least two normalized dynamic parameters is chosen, of which a first describes at least stylus position in dependence on time, so that strokewise a verification measure is obtained indicating similarity of the sample character string to the other sample character string; and wherein for the reference set REF={REF 1 , . . . , REF n } a form matrix of the form: ##EQU8## and a motion matrix of the form: ##EQU9## are formed for all REF i with REF j , i,j=1 . . . n with ##EQU10## as well as ##EQU11## for I=2 . . . I(K), j=2 . . . J(K) as a calculating rule for a verification measure within a stroke k, k=1 . . . K, and ##EQU12## for I=2 . . . I(K), j=2 . . . J(K) as a calculating rule for a verification measure at stroke borders with: ##EQU13## from which a following minimum is further formed: MIN,index!=min (sum((Form.*Form).*(Bewegung.*Bewegung))) where ".*" denotes element-element multiplication whereby the reference signature occurs: Ref={O index }.
Claim 2
Independent Claim
417 words
A method for determining at least one reference autograph character string as a reference set based on a sample set of sample autograph character strings from a same writer, comprising the steps of: through a writing by hand of a predetermined text, producing at least three sample character strings by a same writer for formation of the sample set; during the production of the sample character strings using a stylus, recording and retaining characteristic dynamic parameters describing motion of the stylus; on the basis of the dynamic parameters, and in dependence on whether the stylus is pressed against a writing surface or not, decomposing each sample character string into at least a first and a second partial character string in the form of a first and a second stroke; normalizing the dynamic parameters of all the character strings; determining and retaining differences in the dynamic parameters between a respective sample character string and all sample character strings from the sample set not as yet taken into account in the determination for each dynamic parameter as a parameter deviation; carrying out the determining and retaining of the differences in the dynamic parameters strokewise such that strokes are compared via dynamic programming; determining that sample character string as said at least one reference character string for the reference set which comprises a smallest parameter deviation from all other sample character strings with respect to all dynamic parameters; during calculation of minimal distance between respective strokes, a combination of at least two normalized dynamic parameters is chosen, of which a first describes at least stylus position in dependence on time, so that strokewise a verification measure is obtained indicating similarity of the sample character string to the other sample character string; and wherein for the reference set REF={REF 1 , . . . , REF n } a form matrix of the form: ##EQU8## and a motion matrix of the form: ##EQU9## are formed for all REF i with REF j , i,j=1 . . . n with ##EQU10## as well as ##EQU11## for I=2 . . . I(K), j=2 . . . J(K) as a calculating rule for a verification measure within a stroke k, k=1 . . . K, and ##EQU12## for I=2 . . . I(K), j=2 . . . J(K) as a calculating rule for a verification measure at stroke borders with: ##EQU13## from which a following minimum is further formed: MIN,index!=min (sum((Form.*Form).*(Bewegung.*Bewegung))) where ".*" denotes element-element multiplication whereby the reference signature occurs: Ref={O index }.
Claim 3
Dependent Claim
29 words
A method according to claim 1 in which at least one of the at least two normalized dynamic parameters which are combined is multiplicatively weighted with a weighting factor.
Claim 4
Dependent Claim
21 words
A method according to claim 1 in which one of the normalized dynamic parameters being combined is pressure used in writing.
Claim 5
Dependent Claim
19 words
A method according to claim 1 in which one of the normalized dynamic parameters being combined is writing speed.
Claim 6
Dependent Claim
71 words
A method according to claim 1, in which for the reference set it is investigated with what degree of precision the dynamic parameters are reproduced in relation to individual strokes of the set, and in which a weighting factor is assigned per stroke in dependence on accuracy of reproduction in such a manner that strokes having great reproductive accuracy are strongly weighted and strokes having little reproductive accuracy are weakly weighted.
Claim 7
Dependent Claim
34 words
A method according to claim 5 in which a square of a verification measure is divided by a square of a standard deviation of a stroke being examined in relation to the reference set.
Claim 8
Dependent Claim
46 words
A method according to claim 6, in which the verification measure is divided by the standard deviation of the stroke being examined, in relation to the reference set, multiplied by a mean value of the verification measure formed from the corresponding stroke within the reference set.
Claim 9
Dependent Claim
181 words
A method according to claim 1 for determining two reference signatures from the reference set with REF index =REF 1 as a desired first reference signature and: O Min ={O i }, Iε 1, . . . , n! a partial set of sample signatures having small values in relation to O ref1 , as well as O Max ={O i }, Iε 1, . . . , n! a partial set of sample signatures having large values in relation to O Ref1 ., O Mean ={O i }, Iε 1, . . . , n! a partial set of sample signatures having medium values in relation to O Ref1 . and wherein a following condition holds: ##EQU14## it then follows that MIN,index2!=min, and (sum ((Form.*Form).*(Bewegung.*Bewegung))!.sub.(index1,|OMax |)) with ((Form.*Form).*(Bewegung.*Bewegung))!.sub.(index1,|OMax |) a subcolumn belonging to || OMax || in relation to O Ref1 , and index2 is an index of that of the second reference signature that is most similar to all the ones from this group, and wherein O Ref2 =O index2 , and thus Ref={O Ref1 , O Ref2 }.