JPH0531795B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a continuous character recognition device that recognizes one or more consecutively written characters from the handwriting of input characters.

(Prior Art) Conventionally, a device that recognizes a string of consecutively written characters without the user instructing the end of one character includes, for example, the Japanese Patent Application No. 60-110743.
There was an online continuous character recognition device described in "No. 1" (hereinafter referred to as Reference (1)). Hereinafter, a conventional continuous character recognition device will be explained using this device as an example.

This device uses continuously written characters, including those written without breaks, as an input pattern, and recognizes the characters by pattern matching this pattern with a standard pattern stored in the recognition device in advance. . As a method for recognizing characters by pattern matching, for example, the Institute of Electronics and Communication Engineers Technical Research Report PRL83-29 (September 1983)
On pages 1 to 8 of the 27th) “Stack DP
There is a method as shown in a paper entitled "Online Handwritten Character Recognition by Matching" (hereinafter referred to as Document (2)).

In this method, the input character is first created using a time-series pattern of direction angles of the line segments that make up the character A=(ai
|1≦i≦I). ai represents the direction angle, and I represents the number of line segments. Let this time-series pattern be input pattern A. The standard pattern is stored in advance in the recognition device as a time-series pattern with the same direction angle as the input pattern. This is expressed as a standard pattern Bk=(bj|1≦j≦Jk). k
(1≦k≦K) represents the standard pattern category, bj represents the direction angle, and Jk represents the number of line segments of the standard pattern of category k. i-th data of input pattern A,
The distance between the j-th data of standard pattern Bk is
Defined as the angle between the direction angles ai and bj. Let this distance be d(i,j). The inter-pattern distance Dk between the input pattern A and the standard pattern Bk is a value obtained by accumulating the distances d(i, j) by aligning the time axes of the time series. This time axis matching can be performed by the DP matching method described in document (2). Inter-pattern distance determined in this way
The category with the minimum Dk is taken as the recognition result.

Standard pattern Bk to recognize consecutive characters
This is a character pattern that includes inter-character strokes from the starting point of a character to the starting point of the next character. A continuous character pattern and an isolated character pattern, which is a character pattern written in isolation without including strokes between characters, are prepared.
After concatenating 0 or more consecutive character patterns,
By concatenating single isolated character patterns, a standard pattern of continuous characters is synthesized and used. As a method for efficiently recognizing input patterns expressed in chronological order by concatenating a plurality of standard patterns in a predetermined order, for example,
There is a system used in the continuous speech recognition device described in No. 104669 or Japanese Patent Application No. 55-83199 (hereinafter referred to as Document (3)). The continuous character recognition device shown in Reference (1) targets continuously written characters instead of continuous speech, character strings instead of word strings, and there is a contradiction in the connection of continuous character patterns and isolated character patterns. A special finite state automaton is used to control this.

This finite state automaton takes two states, an initial state s and a final state f, as shown in FIG. When the continuous character pattern BC is accepted, it transitions from the initial state s to the same initial state s, and the isolated character pattern
When BI is accepted, the process transitions from the initial state s to the final state f and ends. By controlling such a finite state automaton, it is possible to recognize the input pattern of consecutively written characters based on a standard pattern of 0 or more consecutive character patterns followed by one isolated character pattern. be able to.

(Problems to be Solved by the Invention) Conventionally, only one type of inter-character stroke has been prepared for a continuous character pattern prepared as a standard pattern, regardless of the preceding and succeeding characters. it is,
For example, when dealing with numbers, the position of the point at which a character ends (hereinafter referred to as the end point) varies depending on the character, but the point at which the character begins to be written (hereinafter referred to as the start point) is at a nearly constant upper position. It is in. This is because the shape of the stroke between characters is assumed to be constant regardless of the characters before and after it. However, for example, when alphanumeric characters are targeted, the shape of the inter-character strokes changes depending on the characters that follow, as shown in FIG. 3, and the pattern may be different from the prepared continuous character pattern. If the inter-character strokes change due to the combination of characters or the stroke order, the characters become unrecognizable.

An object of the present invention is to provide a continuous character recognition device that solves this problem.

(Means for Solving the Problems) The continuous character recognition device according to the present invention includes an input section that reads the handwriting of written characters as a time-series input pattern, and an isolated An isolated character pattern storage unit that stores character patterns as standard patterns, and a plurality of inter-character strokes that represent handwriting from the end point of one character to the start point of the next character are prepared and connected after the isolated character pattern. a continuous character pattern storage unit that holds a continuous character pattern as a standard pattern, and a recognition unit that recognizes the input pattern based on a pattern in which zero or more continuous character patterns and one isolated character pattern are connected. configured.

(Operation) In the present invention, an isolated character pattern is divided into a plurality of sets depending on the position of the writing start point of the character (characters whose starting points change depending on the order of strokes are also treated as separate characters). Then, after the isolated character pattern, a continuous character pattern is created by connecting intercharacter strokes that continue to the respective starting points.

For example, if the starting points are classified into three types, ``top,''``middle,'' and ``bottom,'' the isolated character patterns are divided into three sets as shown in FIG. The continuous character patterns, which are the isolated character patterns plus inter-character strokes, are divided into 9 sets of 3 types each, as shown in FIG.

That is, a continuous character pattern set whose starting point is "Top" and an ending point "Top" TT A continuous character pattern set whose starting point is "Top" and an ending point "Middle" TM A continuous character pattern set whose starting point is "Top" and an ending point "Bottom" ” Continuous character pattern set and TB A continuous character pattern set where the starting point is “Medium” and the ending point is “Top” MT A continuous character pattern set where the starting point is “Medium” and the ending point is “Medium” MM A continuous character pattern set where the starting point is “Medium” Continuous character pattern set with "Middle" and end point "Bottom" MB Continuous character pattern set with "Bottom" start point and "Top" end point BI Continuous character pattern set with "Bottom" start point and "Middle" end point Pattern set BM A continuous character pattern set with the starting point "below" and the ending point "below" BB A set of 9 consecutive character patterns and an isolated character pattern set with the starting point "above"
There are three isolated character pattern sets: TF, an isolated character pattern set whose starting point is "middle", MF, an isolated character pattern set whose starting point is "below", and BF.
In other words, a continuous character pattern whose number is the square of the number of starting point positions and an isolated character pattern whose number is the number of starting point positions are
One set is possible.

Connect zero or more of these continuous character patterns so that the inter-character strokes are connected without any contradiction, and finally, connect one isolated character pattern without any contradiction between the inter-character strokes to create a standard pattern of continuous characters. Synthesize. Here, concatenation without contradiction means that after a continuous character pattern whose ending point is "top", only continuous character patterns or isolated character patterns whose starting point is "top", not "middle" or "bottom", follow. It means to control. In this way, it is possible to connect and recognize a plurality of standard patterns in a predetermined order by changing the conventional control of the finite state automaton.

An example of control by such a finite state automaton is shown in FIG. This finite state automaton has three initial states t, m, b and one final state f. For example, if a continuous character pattern TT whose starting point is "top" and ending point is "top" is accepted,
When a transition occurs from state t to state t and a continuous character pattern TM with a starting point of "top" and an ending point of "middle" is accepted,
Transition from state t to state m, and the starting point is "top",
When the continuous character pattern TB with the end point "bottom" is accepted, a transition is made from state t to state b. in this way,
Transition is repeated between the three states t, m, and b by each continuous character pattern. Then, when any one of the isolated character patterns TF, MF, and BF is accepted, the process transitions to the final state f and ends.

As described above, continuous characters made up of various combinations of characters can be recognized based on a standard pattern in which the start and end points of continuous character patterns and isolated character patterns are connected without contradiction.

Various other finite state automata can be used as the finite state automaton used in the present invention. For example, by using a finite state automaton as shown in FIG. 7, it is possible to recognize only characters that happen to be consecutive.

The principle of the present invention has been explained above. In the above explanation, the starting point is divided into three, but any number may be used. Alphabet has been used as an example of the character type, but the same applies to other general characters. In addition, although we have described the basic online character recognition method using directional data, there are other ways to recognize written characters such as broken characters, etc. A variety of methods can be used, as described in the publication titled ``Online Handwritten Kanji Recognition Moves Toward Relaxing Restrictions''.

(Example) Examples of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the present invention in which the aforementioned starting point is divided into three.

The character pattern input from tablet 1 is
In the preprocessing unit 2, input pattern A is a time-series pattern of direction angles as shown in document (1).
is converted to Continuous character pattern storage units 31-3
9 contains the aforementioned continuous character patterns TT, TM,
TB, MT, MM, MB, BT, BM, and BB are held, and isolated character pattern storage units 41 to 43
holds the aforementioned isolated character patterns IF, MF, and BF. The recognition unit 5, based on these continuous character patterns and isolated character patterns, follows the finite state automaton shown in FIG.
Recognizes input pattern A and outputs result R. As a result, R becomes a character string. The recognition unit 5 performs recognition operations under the control of such a finite state automaton.
The configuration of is shown in document (2), so the details will be omitted.

(Effects of the Invention) According to the present invention, it is possible to obtain a continuous character recognition device that does not require any special operation by the user to divide characters, and can be used for many character types regardless of the stroke order.

[Brief explanation of drawings]

FIG. 1 is a configuration diagram showing one embodiment of the present invention, and FIG.
The figure shows a finite state automaton, and Figure 3 shows the stroke mode between characters in continuous characters.
Figure 4 is a diagram showing the classification of starting points of isolated characters, Figure 5 is a diagram showing continuous character patterns classified by combining the starting points of isolated characters and inter-character strokes, and Figures 6 and 7 are finite states. It is a figure showing an example of control by an automaton.

Claims (1)

[Claims] 1. An input section that reads the handwriting of a character to be written as a time-series input pattern, an isolated character pattern storage section that stores an isolated character pattern representing the handwriting from the beginning to the end of writing as a standard pattern, and a a continuous character pattern storage unit that stores a continuous character pattern as a standard pattern, which prepares a plurality of inter-character strokes representing handwriting from the ending point to the starting point of the next character and connects them after the isolated character pattern; A continuous character recognition device comprising: a recognition unit that recognizes the input pattern based on a pattern in which one or more continuous character patterns and one isolated character pattern are connected.