Method of grouping handwritten word segments in handwritten document images
AssigneeXerox Corporation
InventorTanveer F. Syeda-Mahmood
Life of this patent
Estimating the term…
Abstract
A method and system of recognizing handwritten words in scanned documents, wherein by processing a document containing handwriting, features for word localization are extracted from handwritten words contained in said document through basis points taken from a single curve of text lines. The method is independent of page orientation, and does not assume that the individual lines of handwritten text are parallel, and the method does not require that word regions be aligned with text line orientation wherein intra-word statistics are derived from sample pages rather than using a fixed threshold. The method has applications in digital libraries, handwriting tokenization, document management and OCR systems.
Claims14
A method of grouping scanned handwritten word text line segments to generate handwritten words for the recognition and indexing of documents, said method comprising the steps of:
- generating at least one connected component region of said handwritten word text segments wherein centroid and dominant orientation are identified, said centroid being the first order moments of said region, and said dominant orientation being determined by the direction of the moment of inertia axis of a region;
- selecting peaks in a histogram of said dominant orientation which represent major word orientations in said handwritten word;
- conducting distance computation of said connected component regions through along-line inter-region determination;
- determining curve and corner features from said connected regions by a plurality of boundary points on connected component regions;
- and
- developing intra-word text segment groupings.
A method of recognizing handwritten words in scanned documents, comprising the steps of:
- processing a document containing handwritten words wherein features for word localization are extracted in terms of orientation, skew and intra-word separation from handwritten words contained in said document through basis points taken from a single curve of text lines, and wherein affine coordinates are computed for all features on all curves in a curve group;
- storing said features in a memory;
- accessing said features from memory for comparison to handwritten words in a scanned document to recognize words within said scanned document;
- and
- grouping text segments of handwritten words for purposes of indexing said documents.
A system for recognizing handwritten words by pre-processing a scanned document containing handwriting where features for word localization are extracted from handwritten words contained in said document, said system comprising:
- i) a feature extraction module which forms connecting component regions of a scanned document image representing said handwritten words of said document;
- ii) a curve group generation module assembles groups of curves separated by intra-word separation curve segments belonging to said handwritten words within said scanned document image wherein text lines are independent of page orientation;
- and
- iii) a memory wherein said handwritten words are indexed based on features for word localization extracted from said handwritten words contained in said document and wherein text segments are grouped from said handwritten words for purposes of indexing said documents based on words queries.
A system for the recognition of handwritten words wherein features of handwritten word regions are extracted from said handwritten words for subsequent feature generation, extraction and recognition applications, comprising a microprocessor programmed to:
- generate at least one connected region for each text line of said handwritten words wherein centroid and dominant orientation are identified, said centroid being the first order moments of said region, and said dominant orientation being determined by the direction of the moment of inertia axis of a region;
- select peaks in a histogram of said dominant orientation which represent major word orientations in said handwritten word;
- conduct distance computations of said connected regions through along-line inter-region determination;
- determine curve and corner features from said connected regions by a plurality of boundary points on connected component regions;
- develop intra-word text segment groupings;
- wherein said intra-word text segment groupings are the groupings of text segments into handwritten words that requires the knowledge of intra-word separations between text segments that lie along a line of text, text line detection is independent of page orientation and does not assume that the individual lines of handwritten text are parallel, and all word regions are not required to be aligned with the text line orientation;
- indexing said handwritten words based on features for word localization extracted from said handwritten words contained in said document;
- and
- grouping text segments from said handwritten words for purposes of indexing said documents based on words queries.
Dependencies are inferred from the claim text.