Back to results
US 5,859,929 AGranted
System for character preserving guidelines removal in optically scanned text
AssigneeUnited Parcel Service of America, Inc.
InventorsJing Zhou, Yang He
Life of this patent
Estimating the term…
Abstract
An efficient system and method for reliably identifying guidelines, ruled lines, and the like, in images of text, and distinguishing those portions of such lines which touch or intersect character strokes in the image. The system provides for removal of guideline segments between character strokes without deleting character strokes. The system operates effectively on both machine printed and hand formed text. An image of text has most characters separated for more effective subsequent OCR processing, despite the presence of guidelines connecting the characters.
Claims25
A method of identifying a line pattern in an image of text, comprising the steps of:
- locating contours of isolated objects in said image;
- locating line contour pieces in said contours comprising contiguous pixels forming lines of at least a predetermined straightness along at least a first predetermined length;
- designating clusters of said line contour pieces which have at least a first predetermined closeness along a longitudinal direction generally along the length of said line contour pieces and essentially no transverse overlap;
- locating pairs of said clusters which are spaced apart transversely and have at least a second predetermined closeness to one another generally along a direction transverse to said longitudinal direction;
- and
- eliminating from said clusters those contour pieces that intersect character strokes by:
- finding intersection angles between a contour piece and strokes that connect to it at its ends;
- and
- designating said contour piece part of a line pattern or part of a character responsive to predetermined relationships between said intersection angles, the length of the contour piece, and the distance from the contour piece to the cluster to which it belongs.
6 more elements
A method of removing a line pattern from an image of text, comprising the steps of:
- locating contours of isolated objects in said image;
- locating line contour pieces in said contours comprising contiguous pixels forming lines of at least a predetermined straightness along at least a first predetermined length;
- designating clusters of said line contour pieces which have at least a first predetermined closeness along a longitudinal direction generally along the length of said line contour pieces and essentially no transverse overlap;
- locating pairs of said clusters which are spaced apart transversely and have at least a second predetermined closeness to one another generally along a direction transverse to said longitudinal direction;
- eliminating from said clusters those contour pieces that intersect character strokes by:
- finding intersection angles between a contour piece and strokes that connect to it at its ends;
- designating said contour piece part of a line pattern or part of a character responsive to predetermined relationships between said intersection angles, the length of the contour piece, and the distance from the contour piece to the cluster to which it belongs;
- and
- removing from the image pixels bounded by said paired clusters.
6 more elements
A system for identifying a line pattern in an image of text, comprising:
- an imager operative to acquire and store an image of said text;
- and
- a processor configured to:
- locate contours of isolated objects in said image;
- locate line contour pieces in said contours comprising contiguous pixels forming lines of at least a predetermined straightness along at least a first predetermined length;
- designate clusters of said line contour pieces which have at least a first predetermined closeness along a longitudinal direction generally along the length of said line contour pieces and essentially no transverse overlap;
- locate pairs of said clusters which are spaced apart transversely and have at least a second predetermined closeness to one another generally along a direction transverse to said longitudinal direction;
- and
- eliminate from said clusters those contour pieces that intersect character strokes by:
- finding intersection angles between a contour piece and strokes that connect to it at its ends;
- and
- designating said contour piece part of a line pattern or part of a character responsive to predetermined relationships between said intersection angles, the length of the contour piece, and the distance from the contour piece to the cluster to which it belongs.
9 more elements
A system for removing a line pattern from an image of text, comprising:
- an imager operative to acquire and store an image of said text;
- and
- a processor configured to:
- locate contours of isolated objects in said image;
- locate line contour pieces in said contours comprising contiguous pixels forming lines of at least a predetermined straightness along at least a first predetermined length;
- designate clusters of said line contour pieces which have at least a first predetermined closeness along a longitudinal direction generally along the length of said line contour pieces and essentially no transverse overlap;
- locate pairs of said clusters which are spaced apart transversely and have at least a second predetermined closeness to one another generally along a direction transverse to said longitudinal direction;
- eliminate from said clusters those contour pieces that intersect character strokes, by:
- finding intersection angles between a contour piece and strokes that connect to it at its ends;
- designating said contour piece part of a line pattern or part of a character responsive to predetermined relationships between said intersection angles, the length of the contour piece, and the distance from the contour piece to the cluster to which it belongs;
- and
- remove from the image pixels bounded by said paired clusters.
9 more elements
A method of identifying a line pattern in an image of text, comprising the steps of:
- locating contours of isolated objects in said image;
- locating line contour pieces in said contours comprising contiguous pixels forming lines of at least a predetermined straightness along at least a first predetermined length;
- designating clusters of said line contour pieces which have at least a first predetermined closeness along a longitudinal direction generally along the length of said line contour pieces and essentially no transverse overlap;
- locating pairs of said clusters which are spaced apart transversely and have at least a second predetermined closeness to one another generally along a direction transverse to said longitudinal direction, the closeness of said clusters along said transverse direction being determined by:
- in the case of a first cluster completely transversely overlapping a second cluster, averaging the distances from endpoints of said cluster to a fitted centerline of said first cluster;
- and
- in the case of said first and said second clusters partly transversely overlapping one another, averaging the distances from overlapped endpoints of each of said clusters to a fitted centerline of the other cluster;
- and
- eliminating from said clusters those contour pieces that intersect character strokes.
6 more elements
A method of identifying a line pattern in an image of text, comprising the steps of:
- locating contours of isolated objects in said image;
- locating line contour pieces in said contours comprising contiguous pixels forming lines of at least a predetermined straightness along at least a first predetermined length, said step of locating said line contour pieces in said contours comprising determining pieces of said chain code in which successive short segments form angles with said longitudinal direction less than a predetermined line piece angle threshold;
- designating clusters of said line contour pieces which have at least a first predetermined closeness along a longitudinal direction generally along the length of said line contour pieces and essentially no transverse overlap;
- locating pairs of said clusters which are spaced apart transversely and have at least a second predetermined closeness to one another generally along a direction transverse to said longitudinal direction;
- designating as defining a line pattern those pairs of clusters spaced apart transversely no more than said second predetermined closeness that comprise oppositely directed pieces of chain code and transversely overlap by at least a predetermined proportion;
- and
- eliminating from said clusters those contour pieces that intersect character strokes.
4 more elements
Dependencies are inferred from the claim text.