System for recognizing handwritten words of cursive script
First Claim
1. A method for recognizing a static handwritten word of cursive script, comprising:
- optically reading said word and forming a bit map of pixels representing said word;
skeletonizing said word within said bit map;
segmenting said skeletonized word into one or more primitives, said skeletonized word including a plurality of feature points and said primitives each comprising a continuous segment of said skeletonized word extending between an original feature point and a terminal feature point;
forming a sequence representing the order in which said primitives were written by ordering said primitives in succession beginning at the left side of said word; and
classifying said word by comparing said primitives and said sequence with stored primitives and their associated sequences for known words, wherein the step of forming a sequence comprises;
locating a primitive which is left-most in said word, examining said left-most primitive for the presence of one or more of said end points and designating said left-most primitive as a first primitive if it contains one or more of said end points;
if said left-most primitive does not contain one or more of said end points, examining a primitive connected with said left-most primitive for the presence of an end point, and designating said connected primitive as said first primitive if it contains an end point and designating said left-most primitive as said first primitive if said connected primitive does not contain an end point; and
ordinally designating as subsequent primitives each of said primitives which are connected which are connected with said first primitive and with said subsequent primitives.
0 Assignments
0 Petitions
Accused Products
Abstract
A system for recognizing handwritten words of cursive script by extracting temporal information from static text is described. The temporal information consists of the shape, location and temporal sequence of the segments, or "primitives", constituting the word and is translated from bit maps images of the words into digital information format. By comparing the digital information with formatted digital information for a plurality of known words, as may be stored in computer memory, a list of words having a relatively high probability of matching the static text is generated, and used for word recognition by a process of selecting the most likely match.
44 Citations
31 Claims
-
1. A method for recognizing a static handwritten word of cursive script, comprising:
-
optically reading said word and forming a bit map of pixels representing said word; skeletonizing said word within said bit map; segmenting said skeletonized word into one or more primitives, said skeletonized word including a plurality of feature points and said primitives each comprising a continuous segment of said skeletonized word extending between an original feature point and a terminal feature point; forming a sequence representing the order in which said primitives were written by ordering said primitives in succession beginning at the left side of said word; and classifying said word by comparing said primitives and said sequence with stored primitives and their associated sequences for known words, wherein the step of forming a sequence comprises; locating a primitive which is left-most in said word, examining said left-most primitive for the presence of one or more of said end points and designating said left-most primitive as a first primitive if it contains one or more of said end points; if said left-most primitive does not contain one or more of said end points, examining a primitive connected with said left-most primitive for the presence of an end point, and designating said connected primitive as said first primitive if it contains an end point and designating said left-most primitive as said first primitive if said connected primitive does not contain an end point; and ordinally designating as subsequent primitives each of said primitives which are connected which are connected with said first primitive and with said subsequent primitives. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
-
-
16. Apparatus for recognizing a static handwritten word of cursive script, comprising;
-
means for reading said word and forming a bit map of pixels representing said word; means for skeletonizing said word within said bit map; means for segmenting said skeletonized word into one or more primitives, said skeletonized word including a plurality of feature points and said primitives each comprising a continuous segment of said skeletonized word extending between an original feature point and a terminal feature point; means for forming a sequence representing the order in which said primitives were written by ordering said primitives in succession beginning at the left side of said word; and means for classifying said word by comparing said primitives and said sequence with each of a plurality of stored primitives and their associated sequences for known words, wherein said means for forming a sequence comprises; means for locating a primitive which is left-most in said word, examining said left-most primitive for the presence of one or more of said end points and designating said left-most primitive as a first primitive if it contains one or more of said end points; means for examining a primitive connected with said left-most primitive for the presence of an end point if said left-most primitive does not contain one or more of said end points, and for designating said connected primitive as said first primitive if it contains an end point and designating said left-most primitive as said first primitive if said connected primitive does not contain an end point; and means for ordinally designating as subsequent primitives each of said primitives which are connected with said first primitive and with said subsequent primitive. - View Dependent Claims (17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30)
-
-
31. A system for reading and sorting mail, comprising:
-
means for optically reading an address from a mail piece, said address comprising at least one static word; means for forming a bit map of pixels representing said word; means for skeletonizing said word within said bit map; means for segmenting said skeletonized word into one or more primitives, said skeletonized word including a plurality of feature points and said primitives each comprising a continuous segment of said skeletonized word extending between an original feature point and a terminal feature point and wherein said word comprises two or more connected characters and one or more of said primitives extends between two of the characters; means for forming a sequence representing the order in which said primitives were written by ordering said primitives in succession beginning at the left side of said word; means for classifying said word by comparing said primitives and said sequence with each of a plurality of stored primitives and their associated sequences for known words; and means for directing said mail piece in response to the result of the classification of said word, wherein said means for forming a sequence comprises; means for locating a primitive which is the left-most in said word, examining said left-most primitive for the presence of one or more of said end points and designating said left-most primitive as a first primitive if it contains one or more of said end points means for examining a primitive connected with said left-most primitive for the presence of an end point if said left-most primitive does not contain one or more of said end points, and for designating said connected primitive as said first primitive if it contains an end point and designating said left-most primitive as said first primitive if said connected primitive does not contain an end point; and means for ordinally designating as subsequent primitives each of said primitives which are connected with said first primitive and with said subsequent primitives.
-
Specification