Chinese/Kanji on-line recognition system

• US 4,365,235 A
• Filed: 12/31/1980
• Issued: 12/21/1982
• Est. Priority Date: 12/31/1980
• Status: Expired due to Term

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Greanias, Evon C., Yhap, Ernesto F.
• Primary Examiner(s): Chin, Gary
• Application Number:
  US06/221,863
• Time in Patent Office: 720 Days
• Field of Search: 364/515, 364/518, 364/523, 364/419, 364/200, 364/900, 340/146.3 SY, 340/146.3 AC, 340/751, 178/30, 400/110, 400/484
• US Class Current: 382/189
• CPC Class Codes: G06V 10/20 Image preprocessing
• CPC Class Codes: G06V 10/40 Extraction of image or video features

Abstract

A Chinese/Kanji on-line recognition system consisting of four main sections, these being tablet electronics, a signal filter and segment integration unit, a base stroke classification unit and a symbol element recognition unit, and a symbol recognition output table. The tablet electronics provides pen coordinate signals and pen up/down signals which are applied to the signal filter and segment integration unit to define segments of strokes which correspond to continuous motion of a pen on a tablet in a fixed direction. The base stroke classification unit classifies the motion of the pen between pen down and pen up occurrences into one of 42 categories and also indicates if the stroke has crossed a prior stroke. This is then analyzed by the symbol element recognition unit which interprets the base strokes that have been recorded for the word and generates a sequence of symbol elements, referred to as "alphabet" components, that occur in this symbol. The sequence of symbol elements are interpreted in the symbol recognition output table which provides a form of a simple table look-up to determine the word that had been written. Only 72 basic symbol elements (alphabet) are required to synthesize all of the Chinese/Kanji vocabulary.

Claims

1. An on-line Chinese symbol recognition system with an extendable vocabulary comprising:

an on-line tablet device that collects pen position signals as the Chinese symbol is written,

signal filter and segment integration means for receiving pen coordinate signals and pen up/down signals from said tablet device to define segments of strokes which correspond to continuous motion of a pen on a tablet in a fixed direction,

base stroke classifying means, connected to the output of said signal filter and segment integration means, to classify the motion of the pen between pen down and pen up occurrences into one of a predetermined number of base stroke categories, including means for determining if the stroke has crossed a prior stroke,

symbol element recognition means, connected to the output of said base stroke classification unit, for analyzing the base stroke and crossing information to detect the occurrence of members of a predefined set of symbol elements, each of said symbol elements being graphically descriptive of portions of the shapes of Chinese symbols,

a symbol recognition output table, connected to the output of said symbol element recognition means, which contains a list of all the Chinese symbols in the system'"'"'s vocabulary, each symbol in the list having associated with it all of the symbol element sequences that correspond to it, and

recognition table readout means which accepts the sequence of symbol elements for a given input symbol, searches the recognition output table to determine the identity of the symbol that is associated with that sequence of symbol elements, and sends the symbol identity code to the desired output device.

2. System as recited in claim 1, wherein said signal filter and line segment integration means include means for comparing current pen position data with previous, stored pen position data to thereby filter out excessive data due to pauses in the pen motion or due to random signal fluctuations, and a pause signal detector connected to said comparing means for detecting pause conditions considered in determining segments of strokes.

3. System as recited in claim 1, wherein said signal filter and line segment integration means includes means for determining the beginning and end points of each stroke, means for determining both the average and instantaneous line directions in a line segment for classifying each line segment into compass directions, and storage means for storing the end point coordinates and direction classification of each line segment for subsequent use in said base stroke classifying means.

4. System as recited in claim 3, wherein said means for determining both the average and instantaneous line directions in a line segment includes means for subtracting current point coordinates of a line segment from previous point coordinates.

5. System as recited in claim 1, wherein said base stroke classifying means includes logic means for classifying the line strokes based on the shape of each stroke, the position of each stroke and the intersection of each stroke with other strokes.

6. System as recited in claim 5, wherein said logic means includes stroke logic trees for classifying both single segment strokes and multi-segment strokes.

7. System as recited in claim 1, wherein said symbol element recognition means includes symbol position logic which analyzes the relative position of detected symbol elements and provides the logical representation for the entire symbol by determining the codes for the symbol elements and their sequence of presentation.

8. System as recited in claim 1, wherein said symbol recognition output table contains said list of all the Chinese symbols that are in the system'"'"'s vocabulary, each symbol in the list having associated with it all of the symbol element sequences that correspond to it, and said list of Chinese symbols is extendable by adding to the table new symbols with their logical representation sequences of graphically descriptive symbol elements.

9. System as recited in claim 1, wherein said recognition table readout means includes a display of symbols representing the symbol elements in their approximate relative locations within the word symbol in order to provide the user with valuable feedback regarding the system'"'"'s evaluation of his writing, and provides the user with means for displaying words that have not been included in the system'"'"'s formal vocabulary.

10. System as recited in claim 1, wherein said recognition table readout means includes means for continuously enlarging the vocabulary of recognizable words by adding the codes for symbol elements to the symbol recognition output table.