GB2175426A

GB2175426A - A real time input device for handwritten chinese characters

Info

Publication number: GB2175426A
Application number: GB08607784A
Authority: GB
Inventors: Yiping Jin
Original assignee: Individual
Current assignee: Individual
Priority date: 1985-04-01
Filing date: 1986-03-27
Publication date: 1986-11-26
Also published as: JPS6379126A; GB8607784D0; CN85102853A; CN85102853B

Abstract

A real time input device for handwritten chinese characters comprises a matrix of mutually perpendicular and insulated row and column conductors and a pen-like object with a conducting head. When the pen head touches the plane of the matrix, rows and columns are connected. As a character is written the pen head is moved on the matrix and the coordinates thereof are changed accordingly. The stroke can be recognized by distinguishing the direction and the distance of the movement of coordinates. Recognition is carried out using a small amount of internal memory, only 16 binary digits for the internal code of each character, the possibility of using different sequences of strokes for some characters, no multi-character code for output. These are achieved because two transformations of the primary codes, and indexing pages by radical codes are employed. <IMAGE>

Description

SPECIFICATION A real time input device for a computer for handwritten Chinese characters and its input method The present invention relates to a real time input device of computer for handwritten chinese characters and its input method.

In recent years, some researches on real time input device of computer for recognizing hand written characters and its input method have been carried on in Japan.

"The Patent Laid Open Application of Japan"luEi 58-1280 introduced a device. It contains several pressure sensors under a writting plate. In terms of the distribution of the forces action on the sensors, the coordinate of a pen head used to write characters on the plate can be worked out.

"The Patent Laid Open Application of Japan"i 58-64581 introduced another device. Its pen head is designed to emit light or supersonic. In the two corners of the plate, two sensors are installed, which may follow the tracks of the light or sound source automatically. The coordinate of the pen head can be calculated according to the rotation angles of two sensors.

The above mentioned devices not only have complex structures but also need circuits, such as AID converters, operational amplifiers, etc. The cost is high and the terms of employing them are rather harsh.

"The Patent Laid Open Application of Japan;0# 57-125474 introduced a device of on line recognizing chinese characters. An input structure is recommended, which makes use of a sensor installed in the pen head to ditect pulse-modulated electromagnetic waves emitted by the plate. The coordinate of the pen head is figured out by using a signal filter. It is said that the obtained coordinate is pretty accurate. Nevertheless, its structure is quite complicated. In particu lar, when a stroke is ended, one has to raise the pen a little higher Otherwise, the movement after the pen will be misidentified with an effective part of the previous stroke.

According to the variation of the ditected coordinate, the stroke may be recognized. 42 different strokes are summarized and 72 symbol elements composed of several strokes are then introduced into the above system. When a computer compares symbol elements of a character written down with a dictionary stored inside the computer, indexed according to symbol ele ments of characters, an object cade of the character can be figured out. But in order to write, say, character " i~' ", five symbol elements are needed to construct an internal code, that is, 40 digits of binary number are required. The more the strokes of a character, the longer the code. Therefore, the amount of the labor of compiling a dictionary will be huge and the required amount of internal memory for the dictionary will be too big also.Besides, it is difficut to distinguish those chinese characters having same strokes and thus having the same code.

An object of the present invention is to create a handwritten chinese characters input device whose structure is simple and function is reliable. Another object is to make the dictionary with huge amount of chinese characters can be stored in the computer with limited amount of internal memory and the output should give different codes to different characters.

The flow chart of this invention is shown in Fig. 1. When the pen (2) touches the plate (1), a node of one column and one row on the plate is connected. The sequent numbers of the row and the column are considered as the instantaneous coordinate of the pen. Program (3) is used to trace the coordinate of the pen and to distinguish different strokes. When a stroke ends and the pen is raised, a primary code of the stroke and the coordinates of characteristic points are then stored. At the same time a stroke counter is added by 1. After a character is completed, program (7) will transform the stored primary codes of the strokes, which have many varieties, into so called type I codes having less varieties.The page of the computer dictionary is indexed according to the codes of first several strokes of a character (usually the first several strokes are just the radical of a character). Program (9) is used to drop these radical codes and transform the remaining codes into so called type II codes having still less varieties. Then an internal code is constructed in terms of defined rules. If there are no other characters having the same code as the one written down, an object code of the character can be found on the page of the dictionary, indexed by the radical codes. Otherwise, a flag code of a subroutine to analyze multi character code will be found.Since different character have different forms, therefore, after analyzing multi-character code, a unique object code of the character will be determined and no multi-character code will appear.

The details of the present invention are as follows: The input device includes a matrix writting plate and a pen like object.

The matrix plate are composed of 16 rows and 16 columns of conductors which are mutually perpendicular and insulated (Fig. 3). The column conductor is a piece of teeth-like conductor with 16 notches on it. The row conductor is a piece of strip-like conductor. The 16 row conductors are embedded in the 16 grooves formed of notches of 16 teeth-like column conducters. The upper surfaces of the strips of row conductors and the convex surfaces of the teeth-like column conductors make up the writting plane of the plate.

The plate can also simply be made of a board with double-side copper foils. The column and row conductors are composed of strips of copper foils on the board. If permitted technically, the number of column and row of the matrix can be increased to raise the resolving power.

One end of the pen-like object is made of conductive material, which plays the role of a pen head.

When the pen head gets in touch with one row and one column or more at the same time, at least one cross point (node) of the row and the column is connected.

By scanning continuously over the plate the connected nodes can be found, which is just the coordinate of the pen head. When a stroke is being written, the coordinate of the pen head changes. A computer is used to distinguish different strokes such as

etc, by analyzing the direction and distance of the movement of the coordinate. After a stroke is finished, if the pen head moves to another direction, a new stroke is obtained. Every stroke is represented by a code. When a stroke has been written, a primary code representing this stroke can be obtained.

The separation of pen head with the plate means a stroke has been completed. The primary code and coordinates of characteristic points of the stroke (including the coordinates of the initial point, finally puint and turning point) are then stored, while the stroke counter is added by 1.

After that, the computer is waiting for the input of next stroke or a signal of "the character has been completed".

In order to distinguish different strokes many varieties of the stroke code are needed. However, this will lead to a longer internal code when compiling the computer dictionary. The amount of internal memory required to store the dictionary is correspondingly very big, One feature of the present invention is to transform the primary code twice. The first one is to transform the Drimarv code into tvDe I code which includes 6 or 7 kinds of stroke code: -. I

When the user sends out the signal "The character has been completed" (for instance, press some key) the computer reads out the stored primary codes of the character and makes the first transfermation. The resultant type I codes are then stored in another unit of memory.

Similar to the traditional dictionary indexed by the radicals of characters another feature of the present invention is to index the computer dictionary by the first one, two or several sequent strokes of a character. For example, all the characters with the first five strokes being -, ) , I 7 ,- successively are put into one page. Then all the characters with radical "tri" are on this page (e.g.

It should be noted that those whose radical is not

but having the same first five sequent strokes

are also included on this page. We will still use the name "radical". However, the difference from the traditional meaning should be understood. The advantages of this method are three-fold.First, the codes of the radical are not required to be included in the internal code of a character.

Therefore the internal codes of a character in the computer dictionary is greatly reduced.

Second, the time spent on looking for internal codes is saved, since the searching region is reduced from several thousand to several ten characters on one page. Third, the strokes of some radicals can be written in different order. For instance, all the characters with first three strokes being , I , ' and , , are put into same page. Then the radical can be written in either order.

The second transformation is to reduce the type I codes into four kinds of code representing

code of a character further, so that more informations about the strokes can be stored in the limited internal internal memories. Besides, different forms can be used for some characters, e.g.

can also be written as

This will certainly give rise to multi-character codes, that is, different characters have same internal code. Fortunately, the primary code has been stored. If confusion happens, the primary code may be used to distinguish multi-character code.

The present invention includes a stroke counter. The number of strokes of each character is automatically recorded while it is being written. Therefore the number of strokes of a character can be part of a internal code. The other parts of the internal code are composed of 6 remaining stroke codes (type II code) after dropping the radical code. If the remaining stroke codes are less than 6, zeroes are added on to make 6.

As soon as the computer works out the internal code, it looks up the corresponding page of the computer dictionary. If the character does not have a multi-character code in that page, the object code can be found directly. Otherwise, a subroutine flag code and the sequent number of the strokes needed are given. The flag leads the program to the entrance address of the subrontine. Subroutines distinguish multi-character code based on vavions evidences: the primary code of a specific stroke, the characteristic points of some one or two strokes or the length or slope of some one or two strokes etc. In anycase a unique object code of the input character is given to output after distinguishing multi-character code.

To sum up, the input device of the present invention is very simple and reliable due to the use of touch-connect mode plate and pen. The amount of the internal memory required is small due to the means of twice transformations of primary codes indexing the page by radical code. The internal code of each character has only 16 digits of binary number. There is no multi-character code in the output. Some characters can be written in different sequences of strokes while still have the same internal code. The capacity of the characters of the present invention can be easily expanded. It is also easy to apply this invention to other languages.

A brief explanation of the attached figures follows: Figure 1 is a flow chart of the main part of the present invention.

Figure 2 is a writing board composed of teeth-like and strip-like conductors.

Figure 3 is the top view of a wiring plate composed of board with double-side copper foils.

Figure 4 is the cross section view of a pen-like object.

Figure 5 is a sketch of the circuit of the interface of the computer and the plate.

Figure 6 is a sketch of twice transfermation of primary codes.

Figure 7 is a flow chart of indexing the pages by the radical code.

Figure 8 is a sketch of coding rules of the internal code.

Figure 9 is a sketch of the contents in the memory units of part of the dictionary.

The following is a preferred embodiment of the present invention. The details of the input device and the method of inputing hand written chinese characters are described via examples.

Fig. 2 is a sketch of a 16X16 matrix writing plate. Row conductors are made of 16 pieces of teeth-like conductors. 16 pieces of strip-like column conductors are embedded in the grooves of the teeth-like conductors. All pieces of conductors are mutually insulated. The writing plane ABCD is composed of upper surfaces of the strip-like conductors and the convex surfaces of the teeth-like conductors.

Fig. 4 is a cross section graph of the pen-like object. It has a replaceable pen head made of conducting rubber.

When the pen head gets in touch with one row and one column or more at the same time, at least one node is connected.

The pen head moves on the plane while a character is being written. The coordinate of the connected node changes accordingly on the plane.

The principle of the interface circuit (Fig. 5) is to set all columns at level ''1'' but one column at level "0". The search for level "0" row. If there is no level "0" row, set the nest column at level "0" and repeat above scanning process. Let's take character " #" as an example. In writing the first stroke of '' ", '', the pen head starts from point P1. The node of the 4th column and the 14th row is then connected. If the input of the 4th column is "0", the output of the 14th row is also "0". The coordinate of the starting point P1 is recorded as (14,4). The instantaneous coordinate of the pen head can be traced by repeated scanning. When the pen head moves down 5 intervals from P1 (14,4) to P2 (9,4), the stroke can be recognized as "1".

However when the pen head continuously moves left down from P2 (9,4) to P3 (4,1), the stroke is recognized as ") ". As soon as this stroke is finished and the pen head is raised and separated from the plate, no "0" level row can be found. Therefore, the primary code 05 of and andthe coordinates of starting point P1, final point P3 are stored in the computer. The second stroke which starts at P4 and ends at P9 can recognized as t in the same way. Its primary code is OB. The character " '' is thus written down by finishing all the strokes. The 6 primary codes of " "awl" " are 05, OB, 02, 09, 01, 01.

The computer carries on the first transformation right after the key meant "the character has been completed" is pressed down. The obtained type I codes are 3, 5, 2, 5, 1, 1, as shown in Fig. 6.

After the first transformation the computer determines the page according to the radical code index. From Fig. 7 it can be seen that the character "# " '' is on the page having radical codes (type I code) 3, 5. The radical address, the number of byte and the number of radical stroke 02H are obtained.

The second transformation is to transform the remaining type I codes (without radical codes) into type II codes. From Fig. 6 the type II codes of " Ji, " are easily read out. They are 2, 0, 1, 1. Because the number is less than 6, 2 zeroes are added on, and the complete codes are 2, 0, 1, 1, 0, 0.

Fig. 8 is the rule of compiling internal codes. The internal codes are composed of 16 digits of binary number. The first 4 digits are a number N related to the number of the strokes, N=total number of stroke-number of radical stroke -4. If N < O, let N=O. If N > 15, let N=15. N=O for character ".AL ". The last 4 digits are the last two codes of the type II codes of the character.

The 8 digits in the middle represent the type II codes of the first 4 strokes of the rest part of the character (without radical). Therefore the internal code of '' , " is finally compiled as 0000100001010000, i.e. 0850H.

Fig. 9 shows that on the corresponding page of the computer dictionary the address of the memory unit with content 0850H is An. If 0850H is not a nulti-character code, the object code will be in the memory unit addressed as An+K. Because characters

have same code 0850H, analyzing this multi-character code is required. Now the content of the unit An+K is E12BH. This number is bigger than DOOOH, so it includes a flag code of a subroutine used for analyzing multi-character code. Flag El leads to the entrance of this subroutine. The function of 2B is to indicate the second primary code of the character should be compared with OBH. If it is less OBH, the object code of the character is in the unit An+K+2 (In Fig. 9, the content of that unit is 4914, i.e. the object code of " i ").The second primary code of " Q " is equal to OBH, so the unit An+K +4 should be looked up and its content is EF32. This again includes another subroutine flag code. The function of this subroutine is to compare the row coordinate of the initial point of the third stroke with that of the final point of the second stroke. If the former one is smaller, then the object code of the character is in the unit An+K+4+2. Otherwise, it is stored in the unit An+K+4+4. The content of the former are is 7424, this is the object code of " t ". The content of the latter one is 4881, the object code of " i ".

As an alternative of this embodiment, the writing plate of the present invention can simply be a board with double-side copper foils (Fig. 3). On first side there are 16 parallel strips of copper foil, taken as row conductors. Between every pair of neighboring rows, there are 16 pieces of small copper foil, parallel to the row conductors. On the second side there are 16 strips of copper foil perpendicular to the row conductors on the other side. Every strip connects the 16 pieces of small copper foil on the other side along the direction perpendicular to the row conductors via metallized holes. The column conductors are thus made up. The first side having row conductors and small pieces of copper foil is just the writing plane.

Claims

1. A real time input device of computer for handwritten chinese characters, characterized in that a pen-like object with pen head made of conducting material gets in touch with a matrix plate having mutually perpendicular and insulated row and column conductors so that one or more nodes of rows and columns are connected, the coordinates representing the connected node change as writing characters.

2. A device according to claim 1, characterized in that a matrix plate is composed of parallel teeth-like conductors and parallel strip-like conductors embedded in the grooves of the teeth-like conductors and insulated with the teeth-like conductors, the writing plane is composed of upper surfaces of the strip-like conductors and convex surfaces of the teeth-like conductors.

3. A device according to claim 1, characterized in that a matrix plate is made of a board of double-side copper foil which has several parallel strips of copper foil as row conductors and several small pieces of copper foil between each pair of strips on the frontage side and several strips of copper foil on the reverse side perpendicular to the row conductors, the strips on the raverse side connect the small pieces of copper foil on the frontage side via metallized holes, so the required columns conductors on the frontage side are formed.

4. A device according to claim 1, characterized in that a pen-like object has a replacable pen head made of conducting rubber.

5. A real time input method of computer for handwritten chinese characters by recognizing strokes of the character written on the plate, characterized in that each character is given an internal code composed of stroke codes and the number of stroke, indexed to a page according to radical stroke codes, distinguished by using subroutine in case of several characters on the same page having same code.

6. A method according to claim 5, characterized in that the radical stroke codes used to index the pages is composed of codes of the first one, two or several sequent strokes of each character.

7. A method according to claim 5, characterized in that the internal code of each character takes all or part of the remaining stroke codes after dropping the radical stroke codes as a constituent.

8. A method according to claim 5, characterized in that the internal code of each character takes the code of the number of stroke of that character as a constituent.

9. A method according to claim 6 or 7, characterized in that one or two transformations transform many varieties of the primary stroke codes into less varieties of fundamental codes of strokes, such as

10. A method according to claim 5, characterized in that the subroutine used in distinguishing multi-character code is programmed according to the primary stroke codes, coordinates of characteristic points, the length or the slopes of same stroke.

11. A real time input device for a computer for handwritten Chinese characters substantially as herein described with reference to, and as illustrated in, the accompanying drawings.

12. A real time input method for a computer for handwritten Chinese characters substantially as any one herein described with reference to the accompanying drawings.