Pattern selecting device capable of selecting favorable candidate patterns

Claim 2

A pattern selecting device for producing an output numeral indicative of one of first through N-th candidate patterns which have first through N-th candidate degrees of similarity relative to an input pattern and are decided by a pattern collating system for collating said input pattern with a plurality of registered patterns to produce, successively in a descending order of said first through said N-th candidate degrees, first through N-th candidate pattern signals representative of said first through said N-th candidate patterns, where N represents an integer which is not less than two and is not greater than said plurality, said pattern selecting device comprising: processing means for processing said first through said N-th candidate pattern signals to produce, when an (n+1)-th candidate pattern signal is processed, where n is variable from 1 towards N, an n-th processed signal related to said (n+1)-th through said N-th candidate degrees; estimating means connected to said processing means for estimating an n-th estimated degree for an n-th candidate pattern with reference to said processed signal to produce an estimated signal representative of said n-th estimated degree; and deciding means connected to said processing means and said estimating means for deciding whether or not a ratio of said n-th estimated degree to said n-th candidate degree is not smaller than a predetermined value, said deciding means thereby producing a numeral indicative of said n-th candidate pattern as said output numeral.

Claim 3

A similar pattern selecting device as claimed in claim 1, wherein said processing means comprises: first memory means for memorizing said first through said N-th candidate pattern signals as first through N-th memorized pattern signals; approximation processing means connected to said first memory means for processing (n+1)-th through said N-th memorized pattern signals to produce an n-th approximation signal representative of an n-th approximation variable related to said (n+1)-th through said N-th candidate degrees; second memory means connected to said approximation processing means for memorizing said n-th approximation signal as an n-th memorized approximation signal; and supplying means for supplying said n-th memorized approximation signal as said n-th processed signal to said estimating means.

Claim 4

A similar pattern selecting device as claimed in claim 2, wherein said decision means comprises: calculating means connected to said first memory means and said estimating means for calculating said ratio to produce a ratio signal representative of said ratio; and output decision means connected to said calculating means for deciding whether or not said ratio is not smaller than said predetermined value, said output decision means producing an output signal indicative of said n-th candidate pattern when said ratio is not smaller than said predetermined value.

Claim 5

A pattern selecting device for producing an output numeral indicative of one of first through N-th candidate patterns which have first through N-th candidate degrees of similarity relative to an input pattern and are decided by a pattern collating system for collating said input pattern with a plurality of registered patterns to produce, successively in a descending order of said first through said N-th candidate degrees, first through N-th candidate pattern signals representative of said first through said N-th candidate patterns, where N represents an integer which is not less than two and is not greater than said plurality, said pattern selecting device comprising: processing means for processing said first through said N-th candidate pattern signals to produce, when an (n+1)-th candidate pattern signal is processed, where n is variable from 1 towards N, an n-th processed signal related to said (n+1)-th through said N-th candidate degrees; estimating means connected to said processing means for estimating an n-th estimated degree for an n-th candidate pattern with reference to said processed signal to produce an estimated signal representative of said n-th estimated degree; and deciding means connected to said processing means and said estimating means for deciding whether or not a weighted mean value of said n-th candidate degree and a ratio of said n-th estimated degree to said n-th candidate degree is not smaller than a predetermined value, said deciding means thereby producing a numeral indicative of said n-th candidate pattern as said output numeral.

Claim 6

A pattern selecting device for producing an output numeral indicative of one of first through N-th candidate patterns which have first through N-th candidate degrees of similarity relative to an input pattern and are decided by a pattern collating system for collating said input pattern with a plurality of registered patterns to produce, successively in a descending order of said first through said N-th candidate degrees, first through N-th candidate pattern signals representative of said first through said N-th candidate patterns, where N represents an integer which is not less than two and is not greater than said plurality, said pattern selecting device comprising: processing means for processing said first through said N-th candidate pattern signals to produce, when an (n+1)-th candidate pattern signal is processed, where n is variable from 1 towards N, an n-th processed signal related to said (n+1)-th through said N-th candidate degrees; estimating means connected to said processing means for estimating an n-th estimated degree for an n-th candidate pattern with reference to said processed signal to produce an estimated signal representative of said n-th estimated degree; and deciding means connected to said processing means and said estimating means for deciding whether or not a weighted mean value of said n-th candidate degree and a ratio of said n-th estimated degree to said n-th candidate degree is not smaller than a predetermined value, said deciding means thereby producing a numeral indicative of said n-th candidate pattern as said output numeral.

Claim 7

A pattern selecting device as claimed in claim 4, wherein said processing means comprises: first memory means for memorizing said first through said N-th candidate pattern signals as first through N-th memorized pattern signals; approximation processing means connected to said first memory means for processing (n+1)-th through said N-th memorized pattern signals to produce an n-th approximation signal representative of an n-th approximation variable related to said (n+1)-th through said N-th candidate degrees; second memory means connected to said approximation processing means for memorizing said n-th approximation signal as an n-th memorized approximation signal; and supplying means for supplying said n-th memorized approximation signal as said n-th processed signal to said estimating means.

Claim 8

A pattern selecting device as claimed in claim 5, wherein said decision means comprises: calculating means connected to said first memory means and said estimating means for calculating said weighted mean value to produce a mean value signal representative of said weighted mean value; output decision means connected to said calculating means for deciding whether or not said weighted mean value is not smaller than said predetermined value, said output decision means producing an output signal indicative of said n-th candidate pattern when said weighted means value is not smaller than said predetermined value.

Claim 9

A pattern selecting device as claimed in claim 6, wherein said weighted mean value is a weighted arithmetical mean value Va, said calculating means calculating said weighted arithmetical mean value by the use of an equation given by: Va=(W1×R(n)+W2×S(n))/(W1+W2), where W1 and W2 represent predetermined factors while R(n) represents a ratio of said n-th estimated degree to said n-th candidate degree, and S(n) represents said n-th candidate degree.

Claim 10

A pattern selecting device as claimed in claim 6, wherein said weighted mean value is a weighted geometrical mean value Vg, said calculating means calculating said weighted geometrical mean value Vg by the use of an equation given by: Vg=(R.sup.W1 (n)×S.sup.W2 (n)).sup.1/(W1+W2), where W1 and W2 represent predetermined factors while R(n) represents a ratio of said n-th estimated degree to said n-th candidate degree, and S(n) represents said n-th candidate degree.