US5361311A — Automated recongition of characters using optical…

Claim 1

Independent Claim 337 words

Method for recognizing characters, comprising the steps of: generating a set of optical elements corresponding to a limited set of pairs of positive and negative correlative functions (PNCFs), according to the following method: selecting a set of test characters representative of the characters to be recognized; digitizing the image of each of the test characters; performing positive and negative correlations of each character as to pattern and relevance with each of a set of stored interim PNCFs, and obtaining a set of at least four correlation values responsive to said correlations of each character with each of the interim PNCFs; evaluating said correlation values for each of said characters to determine whether said character resembles character images represented by said interim PNCFs to within a given degree of accuracy; if so, updating said interim PNCFs according to said performed positive and negative correlations; and if not, storing at least one and up to four additional interim PNCFs responsive to said character for comparison with subsequent characters of said set of test characters; and when each of said set of test characters have been thus correlated with the set of interim PNCFs, employing the finally-updated set of interim PNCFs to generate a set of optical elements, each optical element exhibiting variations in optical density responsive to a corresponding PNCF; optically comparing a character to be recognized with each of said set of optical elements, wherein each said comparison is performed by transmitting a pattern of light corresponding to an image of the character to be recognized through an optical element and detecting the total amount of light transmitted through the optical element, to yield a set of transmission coefficients equal in number to said optical elements; supplying said set of transmission coefficients to a neural network; and employing said neural network to compute weighted sums of the transmission coefficients, wherein weights applied to said transmission coefficients in computation of said weighted sums were determined in accordance with previous comparison of known characters with said set of optical elements.

Claim 2

Dependent Claim 337 words

Method for recognizing characters, comprising the steps of: generating a set of optical elements corresponding to a limited set of pairs of positive and negative correlative functions (PNCFs), according to the following method: selecting a set of test characters representative of the characters to be recognized; digitizing the image of each of the test characters; performing positive and negative correlations of each character as to pattern and relevance with each of a set of stored interim PNCFs, and obtaining a set of at least four correlation values responsive to said correlations of each character with each of the interim PNCFs; evaluating said correlation values for each of said characters to determine whether said character resembles character images represented by said interim PNCFs to within a given degree of accuracy; if so, updating said interim PNCFs according to said performed positive and negative correlations; and if not, storing at least one and up to four additional interim PNCFs responsive to said character for comparison with subsequent characters of said set of test characters; and when each of said set of test characters have been thus correlated with the set of interim PNCFs, employing the finally-updated set of interim PNCFs to generate a set of optical elements, each optical element exhibiting variations in optical density responsive to a corresponding PNCF; optically comparing a character to be recognized with each of said set of optical elements, wherein each said comparison is performed by transmitting a pattern of light corresponding to an image of the character to be recognized through an optical element and detecting the total amount of light transmitted through the optical element, to yield a set of transmission coefficients equal in number to said optical elements; supplying said set of transmission coefficients to a neural network; and employing said neural network to compute weighted sums of the transmission coefficients, wherein weights applied to said transmission coefficients in computation of said weighted sums were determined in accordance with previous comparison of known characters with said set of optical elements.

Claim 3

Dependent Claim 43 words

The method of claim 1, wherein said step of transmitting a pattern of light through an optical medium is performed by digitizing an image of the character and activating individual elements of an array of light emitting devices corresponding to said digitized image.

Claim 4

Dependent Claim 14 words

The method of claim 2, wherein said light emitting devices are light emitting diodes.

Claim 5

Dependent Claim 53 words

The method of claim 2, wherein the complete set of optical elements includes a number of sets each comprising four optical elements, corresponding to four PNCFs, one element of each set exhibiting optical density variation corresponding to positive pattern information, to negative pattern information, to positive relevance information, and to negative relevance information.

Claim 6

Dependent Claim 31 words

The method of claim 4, wherein an inverted image of a character to be recognized is optically compared to said optical elements corresponding to said negative pattern and negative relevance information.

Claim 7

Dependent Claim 38 words

The method of claim 1, wherein said neural network employed for computing weighted sums is a network of resistors connected between nodes of said network, the values of said resistors corresponding to weights for computing said weighted sums.

Claim 8

Dependent Claim 25 words

The method of claim 1, wherein each said character is simultaneously compared to each of the optical elements corresponding to said limited set of PNCFs.

Claim 9

Dependent Claim 32 words

The method of claim 1, wherein said step of detecting the total amount of light transmitted through the optical element is performed employing a photodetector corresponding to each of said optical elements.

Claim 10

Independent Claim 292 words

A method of operating a system for optical character recognition employing optical comparison of characters to be recognized to a limited set of pairs of positive and negative correlative functions (PNCFs), said sets of PNCFs including functions representing both pattern and relevance information, said method comprising a learning mode and an operating mode, said learning mode comprising the steps of: constructing a set of varyingly transparent optical elements, each exhibiting variation in optical density corresponding to one of said limited set of PNCFs; selecting a set of known test characters; transmitting a pattern of light corresponding to each of said set of test characters through each of said set of optical elements and measuring the amount of each pattern of light transmitted through each set of optical elements, thus determining a set of transmission coefficients corresponding to the transmission of the patterns of light corresponding to each of said set of known test characters through each of said optical elements; and analyzing the set of transmission coefficients to establish weights in a neural network for summing transmission coefficients responsive to subsequent transmission of patterns of light corresponding to characters to be recognized through said set of optical elements; and said operating mode comprising the steps of: transmitting a pattern of light corresponding to each of said set of characters to be recognized through each of said set of optical elements while measuring the amount of said pattern of light transmitted through each of said set of optical elements, thus determining a set of transmission coefficients responsive to said measurements; supplying said set of said transmission coefficients corresponding to said character to be recognized to said neural network; and employing said neural network to determine the identity of the character to be recognized.

Claim 11

Independent Claim 292 words

A method of operating a system for optical character recognition employing optical comparison of characters to be recognized to a limited set of pairs of positive and negative correlative functions (PNCFs), said sets of PNCFs including functions representing both pattern and relevance information, said method comprising a learning mode and an operating mode, said learning mode comprising the steps of: constructing a set of varyingly transparent optical elements, each exhibiting variation in optical density corresponding to one of said limited set of PNCFs; selecting a set of known test characters; transmitting a pattern of light corresponding to each of said set of test characters through each of said set of optical elements and measuring the amount of each pattern of light transmitted through each set of optical elements, thus determining a set of transmission coefficients corresponding to the transmission of the patterns of light corresponding to each of said set of known test characters through each of said optical elements; and analyzing the set of transmission coefficients to establish weights in a neural network for summing transmission coefficients responsive to subsequent transmission of patterns of light corresponding to characters to be recognized through said set of optical elements; and said operating mode comprising the steps of: transmitting a pattern of light corresponding to each of said set of characters to be recognized through each of said set of optical elements while measuring the amount of said pattern of light transmitted through each of said set of optical elements, thus determining a set of transmission coefficients responsive to said measurements; supplying said set of said transmission coefficients corresponding to said character to be recognized to said neural network; and employing said neural network to determine the identity of the character to be recognized.

Claim 12

Dependent Claim 47 words

The method of claim 9, wherein said step of transmitting a pattern of light corresponding to a character through an optical element is performed by digitizing an image of the character and activating individual elements of an array of light emitting devices corresponding to said digitized image.

Claim 13

Dependent Claim 14 words

The method of claim 10, wherein said light emitting devices are light emitting diodes.

Claim 14

Dependent Claim 56 words

The method of claim 9, wherein the complete set of optical elements corresponding to said PNCFs includes a number of sets of four elements, one element of each of said sets of four elements exhibiting optical density variation corresponding to positive pattern information, to negative pattern information, to positive relevance information, and to negative relevance information.

Claim 15

Dependent Claim 40 words

The method of claim 9, wherein an inverted image of a character to be recognized is optically compared to said optical representations implemented by said optical elements corresponding to said negative pattern and negative relevance information corresponding to each PNCF.

Claim 16

Dependent Claim 40 words

The method of claim 9, wherein said neural network employed for computing weighted sums is a plurality of resistors connected between nodes of said network, the values of said resistors corresponding to said established weights for comparing said weighted sums.

Claim 17

Dependent Claim 35 words

The method of claim 9, wherein a pattern of light corresponding to a character to be recognized is simultaneously transmitted through each of the optical elements corresponding to one of said limited set of PNCFs.

Claim 18

Dependent Claim 26 words

The method of claim 9, wherein said steps of said learning mode are performed at a time of assembly of said system for optical character recognition.

Claim 19

Dependent Claim 151 words

Apparatus for optical identification of characters, comprising: means for transmitting a pattern of illumination corresponding to the image of an individual character to be identified through a series of optical elements each exhibiting variation in optical density corresponding to one of a predetermined set of pairs of positive and negative correlative functions (PNCFs), said PNCFs including both pattern and relevance information; means for measuring the amount of light transmitted from each of said patterns through each of said corresponding optical elements, and for generating a transmission coefficient responsive to each said measurement; a neural network comprising a winner-take-all circuit for computing weighted sums of the transmission coefficients corresponding to the amount of light transmitted by essentially identical patterns of illumination corresponding to each character to be recognized through each of the optical elements; and means for identifying the character to be identified responsive to said weighted summation of the transmission coefficients.

Claim 20

Dependent Claim 151 words

Apparatus for optical identification of characters, comprising: means for transmitting a pattern of illumination corresponding to the image of an individual character to be identified through a series of optical elements each exhibiting variation in optical density corresponding to one of a predetermined set of pairs of positive and negative correlative functions (PNCFs), said PNCFs including both pattern and relevance information; means for measuring the amount of light transmitted from each of said patterns through each of said corresponding optical elements, and for generating a transmission coefficient responsive to each said measurement; a neural network comprising a winner-take-all circuit for computing weighted sums of the transmission coefficients corresponding to the amount of light transmitted by essentially identical patterns of illumination corresponding to each character to be recognized through each of the optical elements; and means for identifying the character to be identified responsive to said weighted summation of the transmission coefficients.

Claim 21

Dependent Claim 72 words

The apparatus of claim 17, wherein said neural network comprises a network of resistances connected between input nodes, intermediate nodes and output nodes, whereby voltages corresponding to transmission coefficients at said input nodes are weighted by the values of resistors connected between said input nodes, said intermediate nodes, and said output nodes, and whereby the voltage present on the output nodes is indicative of the identity of the character to be identified.

Claim 22

Dependent Claim 38 words

The apparatus of claim 17, wherein said optical elements each comprise a transparent member having a varyingly dense pattern of opaque dots printed thereon to provide varying optical density thereof corresponding to one of said set of PNCFs.

Claim 23

Dependent Claim 195 words

Apparatus for optical identification of characters, comprising: means for transmitting a pattern of illumination corresponding to the image of an individual character to be identified through a series of optical elements each exhibiting variation in optical density corresponding to one of a predetermined set of pairs of positive and negative correlative functions (PNCFs), said PNCFs including both pattern and relevance information; means for measuring the amount of light transmitted from each of said patterns through each of said optical elements, and for generating a transmission coefficient responsive to each said measurement; a neural network for computing weighted sums of the transmission coefficients corresponding to the amount of light transmitted by essentially identical patterns of illumination corresponding to each character to be recognized through each of the optical elements; and means for identifying the character to be identified responsive to said weighted summation of the transmission coefficients, wherein the complete set of optical elements corresponding to each PNCF includes a number of sets, each including four elements, the element of each set including elements exhibiting optical density variation corresponding to positive pattern information, to negative pattern information, to positive relevance information, and to negative relevance information.

Claim 24

Dependent Claim 195 words

Apparatus for optical identification of characters, comprising: means for transmitting a pattern of illumination corresponding to the image of an individual character to be identified through a series of optical elements each exhibiting variation in optical density corresponding to one of a predetermined set of pairs of positive and negative correlative functions (PNCFs), said PNCFs including both pattern and relevance information; means for measuring the amount of light transmitted from each of said patterns through each of said optical elements, and for generating a transmission coefficient responsive to each said measurement; a neural network for computing weighted sums of the transmission coefficients corresponding to the amount of light transmitted by essentially identical patterns of illumination corresponding to each character to be recognized through each of the optical elements; and means for identifying the character to be identified responsive to said weighted summation of the transmission coefficients, wherein the complete set of optical elements corresponding to each PNCF includes a number of sets, each including four elements, the element of each set including elements exhibiting optical density variation corresponding to positive pattern information, to negative pattern information, to positive relevance information, and to negative relevance information.

Claim 25

Dependent Claim 35 words

The apparatus of claim 17, further comprising means for digitizing an image of said character, and means for controlling said means for generating a pattern of illumination responsive to the digitized image of said character.

Annuity Stage	Window Opens	Payment Deadline	Surcharge Period Until	Estimated Fee	Status
3.5 Years (Year 4)	1997-10-31	1998-05-02	1998-11-01	$2000.00	Window Closed / Lapsed
7.5 Years (Year 8)	2001-10-31	2002-05-02	2002-11-01	$3760.00	Window Closed / Lapsed
11.5 Years (Year 12)	2005-10-31	2006-05-02	2006-11-01	$7700.00	Window Closed / Lapsed

Automated recongition of characters using optical filtering with positive and negative functions encoding pattern and relevance information

Abstract

Inventors

Bibliographic Details

Claims Analysis

References Cited (Backward Citations)

USPTO Annuity & Maintenance Windows

Assigned Categories & Tags

Assign Category Tag