Device and method for pattern recognition

Claim 2

A pattern recognition device comprising: image pickup means for detecting a pattern of a figure to be identified and for producing a corresponding binary coded figure pattern; shape discerning means for projecting said figure pattern on mutually perpendicular X and Y axes, deriving a histogram of said figure for each of said X and Y projections, and first order differentiating said histograms along said X and Y axes to produce a series of characteristic values for said figure pattern, said characteristic values for said figure pattern being defined as a set of numbers for all possible combinations of the array of signs of the positive and negative pulses occurring in the first order differentiation waveform of said histograms; and processing means for comparing said figure pattern values to predetermined values of known figure patterns, to identify said figure to be identified.

Claim 3

The pattern recognition device of claim 1, wherein said shape discerning means further includes means for deriving, through predetermined arithmetic operational formulas, from said characteristic values, a shape matching value for use in said processing means for comparing said figure pattern shape matching value to shape matching values of known figure patterns to identify said figure to be identified.

Claim 4

The pattern recognition device of claim 2, said shape discerning means further including means for counting the number of said positive and negative pulses and posture recognition means responsive to the regularity with which said number of pulses varies in response to predetermined rotation of the posture of a known figure for identifying said figure to be identified.

Claim 5

A pattern recognition method, comprising: detecting the pattern of a figure to be identified and producing a corresponding image signal; producing a binary coded pattern corresponding to said figure; projecting said binary coded pattern onto mutually perpendicular X and Y axes; differentiating the projected pattern to produce X and Y differentiated waveforms; obtaining from said waveforms data characteristic of said binary coded pattern, the step of obtaining data characteristic of said pattern including determining all possible combinations of the array of signs of the positive and negative pulses occurring in said waveforms and providing a set of numbers for said combinations, where each number corresponds to the frequency of occurrence of an array of positive and negative signs in a plurality of neighboring pulses in said waveforms, said set of numbers comprising said characteristic data for identification of said figure; and comparing the data obtained from said waveforms with known data to identify said figure.

Claim 6

A pattern recognition method, comprising: detecting the pattern of a figure to be identified and producing a corresponding image signal; producing a binary coded pattern corresponding to said figure; projecting said binary coded pattern onto mutually perpendicular X and Y axes; differentiating the projected pattern to produce X and Y differentiated waveforms; obtaining from said waveforms data characteristic of said binary coded pattern, the step of obtaining data characteristic of said pattern including determining all possible combinations of the array of signs of the positive and negative pulses occurring in said waveforms and providing a set of numbers for said combinations, where each number corresponds to the frequency of occurrence of an array of positive and negative signs in a plurality of neighboring pulses in said waveforms, said set of numbers comprising said characteristic data for identification of said figure; and comparing the data obtained from said waveforms with known data to identify said figure.

Claim 7

A pattern recognition method, comprising: detecting the pattern of a figure to be identified and producing a corresponding image signal; producing a binary coded pattern corresponding to said figure; projecting said binary coded pattern onto mutually perpendicular X and Y axes; differentiating the projected pattern to produce X and Y differentiated waveforms; obtaining from said waveforms data characteristic of said binary coded pattern, the step of obtaining data characteristic of said pattern including determining all possible combinations of the array of signs of the positive and negative pulses occurring in said waveforms, providing a set of numbers for said combination, where each number corresponds to the frequency of occurrence of an array of positive and negative signs in a plurality of neighboring pulses in said wave forms, and determining from said set of numbers, by means of arithmetic operational formulas, shape selection values, and shape selection values comprising the characteristic data for identification of said figure; and comparing said characteristic data with known data to identify said figure.

Claim 8

A pattern recognition method, comprising: detecting the pattern of a figure to be identified and producing a corresponding image signal; producing a binary coded pattern corresponding to said figure; projecting said binary coded pattern onto mutually perpendicular X and Y axes; differentiating the projected pattern to produce X and Y differentiated waveforms; obtaining from said waveforms data characteristic of said binary coded pattern, the step of obtaining data characteristic of said pattern including determining all possible combinations of the array of signs of the positive and negative pulses occurring in said waveforms, providing a set of numbers for said combination, where each number corresponds to the frequency of occurrence of an array of positive and negative signs in a plurality of neighboring pulses in said wave forms, and determining from said set of numbers, by means of arithmetic operational formulas, shape selection values, and shape selection values comprising the characteristic data for identification of said figure; and comparing said characteristic data with known data to identify said figure.

Claim 9

A pattern recognition device comprising: image pickup means for detecting a pattern of a figure to be identified and for producing a corresponding binary coded figure pattern; shape discerning means for projecting said figure pattern on mutually perpendicular X and Y axes, deriving a histogram of said figure for each of said X and Y projections, and first order differentiating said histograms along said X and Y axes to produce a series of characteristic values for said figure pattern, said characteristic values for said figure pattern being defined as the frequency of occurrence of the predetermined array of the positive and negative signs of a plurality of neighboring pulses derived in the first order differentiation of said histograms, said plurality of neighboring pulses being evaluated for all possible combinations of signs, and said shape discerning means further including means for deriving, through predetermined arithmetic operational formulas, from said characteristic values, a shape matching value; and processing means for comparing said shape matching values to predetermined values of known figure patterns, to identify said figure to be identified.

Claim 10

A pattern recognition device comprising: image pickup means for detecting a pattern of a figure to be identified and for producing a corresponding binary coded figure pattern; shape discerning means for projecting said figure pattern on mutually perpendicular X and Y axes, deriving a histogram of said figure for each of said X and Y projections, and first order differentiating said histograms along said X and Y axes to produce a series of characteristic values for said figure pattern, said characteristic values for said figure pattern being defined as the frequency of occurrence of the predetermined array of the positive and negative signs of a plurality of neighboring pulses derived in the first order differentiation of said histograms, said plurality of neighboring pulses being evaluated for all possible combinations of signs, and said shape discerning means further including means for deriving, through predetermined arithmetic operational formulas, from said characteristic values, a shape matching value; and processing means for comparing said shape matching values to predetermined values of known figure patterns, to identify said figure to be identified.

Claim 11

The pattern recognition device of claim 6, wherein said processing means compares said shape selection values to predetermined shape selection values of known figure patterns to identify said figure to be identified.

Claim 12

The pattern recognition device of claim 6, wherein said predetermined arithmetic operational formulas for deriving the shape matching values (M X *, M Y *) are as follow, M.sub.X *=(M.sub.X.sup.++ √M.sub.X.sup.--)+(M.sub.X.sup.+- ·M.sub.X.sup.-+) M.sub.Y *=(M.sub.Y.sup.++ ·M.sub.Y.sup.--)+(M.sub.X.sup.+-i ·M.sub.Y.sup.-+) where M X ++ , M X +- , M X -+ , and M X -- are characteristic values obtained from the differentiated waveform of the projected figure on the X axis, and M Y ++ , M Y +- , M Y -+ , and M Y -- are characteristic values obtained from the differentiated waveform of the projected figure on the Y axis.

Claim 13

The pattern recognition device of claim 6, wherein said predetermined arithmetic operational formulas for deriving the shape matching value are as follows, M.sup.*.sub.X =(M.sub.X.sup.++ M.sub.X.sup.--)+(M.sup..sub.X +- M.sub.X.sup.-+) M.sub.Y *=(M.sub.Y.sup.++ M.sub.Y.sup.--)+(M.sub.Y.sup.+- M.sub.Y.sup.-+) M*=M.sub.X *+M.sub.Y * where M X ++ , M X +- , M X -+ , and M X -- are characteristic values obtained from the differentiated waveform of the projected figure on the X axis, M Y ++ , M Y +- , M Y -+ , and M Y -- are characteristic values obtained from the differentiated waveform of the projected figure on the Y axis, and M* is the shape selection value.

Claim 14

The pattern recognition device of claim 6, wherein said predetermined arithmetic operational formulas for deriving the shape matching values (M X *, M Y *) are as follow, M.sub.X *=M.sub.X.sup.++ ×1000=M.sub.X.sup.+- ×100+M.sub.X.sup.-+ ×10+M.sub.X.sup.-- M*.sub.Y *=M.sub.Y.sup.++ ×1000=M.sub.Y.sup.+- ×100+M.sub.Y.sup.-+ ×10+M.sub.Y.sup.-- where M X ++ , M X +- , M X -+ , and M X -- are characteristic values obtained from the differentiated waveform of the projected figure on the X axis, and M Y ++ , M Y +- , M Y -+ , and M Y -- are characteristic values obtained from the differentiated waveform of the projected figure on the Y axis.

Claim 15

The pattern recognition device of claim 6, wherein said predetermined arithmetic operational formulas for deriving the shape matching value (M * ) are as follow, M.sub.X *=M.sub.X.sup.++ ×1000=M.sub.X.sup.+- ×100+M.sub.X.sup.-+ ×10+M.sub.X.sup.-- M.sub.Y *=M.sub.Y.sup.++ ×1000=M.sub.Y.sup.+- ×100+M.sub.Y.sup.-+ ×10+M.sub.Y.sup.-- M*=M.sub.X *+M.sub.Y * where M X ++ , M X +- , M X -+ , and M X -- are characteristic values obtained from the differentiated waveform of the projected figure on the X axis, and M Y ++ , M Y +- , M Y -+ , and M Y -- are characteristic values obtained from the differentiated waveform of the projected figure on the Y axis.