VIEWS: 9 PAGES: 1 POSTED ON: 1/14/2010
What should we do for digital handwriting? Zhang xi-wen firstname.lastname@example.org People can easily capture handwriting and drawing and directly enter them into computer using digital pen and paper. Thus, it is necessary to develop a software prototype to capture, store, manage, browse, view, process, modify, recognize and understand handwriting. The report discusses their usage modes and implementation approaches in variable cases. The size, color, grid of paper and the ink color, line width, handwriting type of pen are involved in recording handwriting. The position, timestamp, and pressure of sampling points are used to produce three styles of handwriting. Increment of position, similarity of time and pressure are exploited to reduce storage memory of handwriting. Handwriting files are managed, searched and browsed according to their names, keywords, timestamps and contents. Global and detailed styles are used to view handwriting. Filters are used to delete redundant points and adjust non-accurate points to smooth handwriting. The cubic Bezier and Bspline fitting are used to represent handwriting in smoothed appearance. In writing, user can undo, redo and delete some strokes and points. Translation, rotation, scaling, delete, replace of some strokes and insert strokes are to modify handwriting in free and associated manners. The corresponding handwriting before and after modifications is also marked using different color. The color, width, and style of selected strokes can be edited. Patches are extracted based on compact properties of strokes. Graphics\text separation is performed according to the regularity of stroke block representing text. The multiple levels of structures, including characters (English words), lines, paragraphs, and sections, are extracted from handwriting with automatic clustering algorithms and interaction ways. In character recognition, the original character with strokes and its first candidate character are displayed in the line-to-line way. Its other candidate characters can also be selected from a list. The candidate characters of neighboring original characters are also used to improve character recognition by incorporating a Chinese language model. It is implemented with Dynamic Programming or Genetic Algorithm. The recognition of speech corresponding to the handwriting is used to improve handwriting recognition with some optimization Algorithms. Strokes representing graphics are classified into two types: primitive ones and compound ones. Some compound strokes are recognized as complex graphics. A compound stroke is segmented into multiple primitive strokes. Primitive strokes are classified into: straight line segments, circular arcs, circles, elliptical arcs (regular, tilting), ellipses (regular, tilting), cubic Bezier curves, cubic Bspline curves. The strokes with types are fitted to corresponding graphics. The graphics are represented with some key parameters. Some complex graphics, such as polylines and polygons, are extracted from the primitive graphics. Some text and some graphics are combined to generate an object based on domain knowledge. The software is to convert handwriting into formal objects. It is challenging and promising to advance application of digital handwriting.
Pages to are hidden for
"Digital pen and digital paper (tablet) are available with lower"Please download to view full document