CN1105985C

CN1105985C - Device and method for Chinese input by hand writing and speech sound

Info

Publication number: CN1105985C
Application number: CN 98125323
Authority: CN
Inventors: 薛其祥; 陈淮琰; 吴卫国
Original assignee: Inventec Besta Xian Co Ltd
Current assignee: Yingyeda Co Ltd (Shanghai)
Priority date: 1998-12-11
Filing date: 1998-12-11
Publication date: 2003-04-16
Anticipated expiration: 2018-12-11
Also published as: CN1257240A

Abstract

The present invention relates to a device and a method for Chinese input by handwriting and pronunciation (phonetic notation or phonetic transcription), particularly to a handwriting and pronunciation Chinese input method which can be suitable for a handy computer (HPC). First, handwriting and pronunciation data is obtained according to an input device under the conditions that a large memory is not used for storing data for Chinese identification and a keyboard is not used; second, the handwriting and pronunciation data is sent to a pronunciation identification system by preprocessing and characteristics; third, an identification result is determined by the difference match of the pronunciation identification system; finally, a pronunciation comparison table of Chinese characters of Chinese is found according to identified pronunciation letters, and required Chinese characters of Chinese are obtained to reach an input purpose.

Description

Hand-written pronunciation input in Chinese apparatus and method

The present invention relates to a kind of apparatus and method that realize the Chinese character input on palmtop computer, relates in particular to a kind of handwriting input pronunciation that utilizes, and searches the input method and the device thereof of corresponding Chinese with identification.

The high-tech electronic technology in the world today is with people's speed fast development difficult to the appraisal.Wherein miniaturization and the hand held massaging device that is easy to move rely on characteristics such as its light, super book, exquisiteness, come out top especially in numerous scientific and technological electronic products in today.Palmtop computer (hereinafter to be referred as HPC) promptly is a member that wherein has characteristic.

Present HPC product, the method for inputs such as Chinese hand-writing input method, pronunciation input method Chinese has become its standard configuration.There are not these input methods that the HPC product is in a disadvantageous position in intense market competition.

Please refer to Figure 1 and Figure 2, this two figure illustrates the identification principle of Chinese hand-writing input method.On input media 10, the writing Chinese characters font with the Hanzi font back input identification system core 11 that quantizes, does that pre-service 111, pattern express 112, coupling is differentiated 113 and picked out Chinese character.Its main treatment scheme is to obtain Chinese font data by input media 10 through sampling (step 200), after removing noise, pre-service (201) becomes standardized binary numeral, this binary numeral is carried out Filtering Processing (202) and extracts certain pattern and express that (feature extraction or primitive are cut apart identification, here so-called primitive is cut apart, be by binary numeral is scanned in coordinate system, and be divided into one by one independently primitive.Each primitive all is that last one group of binary numeral is formed, and in coordinate system, these numerical value usable directions and length are represented.With these directions and length data as feature extraction, for the primitive identification.) form (203), earlier be stored in standard Chinese word radical expression-form coupling known in the radical data and differentiate (204), again be stored in after standard Chinese word table known in the dictionary 114 (feature templates or syntactic rule set) reaches form and mate differentiations (classifying or syntactic analysis) (205), just can identify the unknown Chinese character of being imported (206).

Consult initial consonant and the final key position arrangement plan (as shown in Figure 3) and the Chinese character pronunciation table of comparisons (as shown in Figure 4) of keyboard, the process flow diagram that cooperates Fig. 5 again, pronunciation input method is based on Chinese pronunciation, read key (300) by keyboard, English letter of keyboard is converted to pronunciation letter (phonetic notation or phonetic) (301), press the Chinese character pronunciation table of comparisons, find out corresponding Chinese words sequence (302), and this Chinese character sequence is shown that (303) go out for the user select, and reach the purpose that reaches the pronunciation input by the click of English letter of keyboard.

Utilize the mode of Chinese hand-writing input method, pronunciation input method input Chinese character in the known technology, truly have its convenience.But in actual applications, but there is considerable shortcoming in known technology for the application on HPC.

The shortcoming of pronunciation input method:

1, pronunciation input method is that the pronunciation letter is corresponded respectively to the English alphabet of keyboard and imports the pronunciation letter, and is relatively more difficult for the user who is unfamiliar with keyboard.

2, pronunciation input method need be remembered the corresponding tables of English letter of keyboard and pronunciation letter.

3, for reducing hardware cost, and for consideration easy to carry etc., many HPC do not design hardware keyboards, and these manufacturers are for compatibility and user interface (UI), and the soft keyboard of trying every possible means to have realized is to remedy the deficiency of on-keyboard.So-called soft keyboard is meant with software and comes simulating keyboard, and different is that hard manual will knock keyboard input, and soft keyboard can only be pointed out with pen.Compare with the use of soft keyboard, directly write and want much direct.(annotate: soft keyboard is to touch or with the keyboard that shows on the taps screen, general keyboard desire claims hard manual or keyboard with hand.)

The shortcoming of hand-writing input method:

For the hand-written identification system on the HPC, the Chinese hand-written identification system will take suitable space, usually the Chinese hand-written identification system of making about 15000 words approximately need take the 1M space, these all can increase cost, and along with the expansion of character set, identification speed, the discrimination power of identification system can descend to some extent.

If therefore can be under the situation of not using big Identification Data, with hand-written pronunciation input Chinese character, nature can reduce the occupation proportion of system memory space, also can solve the shortcoming of most of prior art.But the prerequisite of this new method is, treatment step must be very simple, just is unlikely to increase the burden of system, and treatment step must fully meet the pattern that general numerical data is handled, so just easier being applied on the HPC field simultaneously.

In view of this, fundamental purpose of the present invention is to provide a kind of apparatus and method that realize input in Chinese, can utilize existing pronunciation input method under the situation that does not need big Identification Data, reaches the purpose of input Chinese character with handwriting sckeme.

According to above-mentioned purpose, the invention provides a kind of hand-written pronunciation Chinese character input method, applicable to products such as HPC.At first utilize the core algorithm of Chinese handwriting identification to make the pronunciation identification system at the character set of tens letters of Chinese pronunciation, only need the space of tens K bytes, then according to the resulting hand-written pronunciation data of input media, give the pronunciation identification system through pre-service and feature extraction, the identification system coupling is differentiated the back and is determined identification result, according to the pronunciation letter searching Chinese character pronunciation table of comparisons that picks out, just can obtain required Chinese words at last again.

Below contrast accompanying drawing,, do one and describe in detail, understand feature of the present invention is had further to structural design of the present invention and principle of operation.Among the figure:

Fig. 1 is the handwriting identification principle schematic;

Fig. 2 is a handwriting identification process flow diagram shown in Figure 1;

Fig. 3 is the initial consonant and the vowel keyboard arrangement plan of keyboard;

Fig. 4 is the Chinese character pronunciation table of comparisons synoptic diagram of keyboard;

Fig. 5 is the process flow diagram of Fig. 3, pronunciation input method shown in Figure 4;

Fig. 6 is the hardware square structural drawing in palmtop computer (HPC) equipment of the embodiment of the invention;

Fig. 7 is the process flow diagram of hand-written pronunciation input method of the present invention;

Fig. 8 is the hand-written phonetic state of hand-written pronunciation input " peace " word example;

Fig. 9 is the identification state of hand-written pronunciation input method " peace " word example;

Figure 10 is the contrast table of hand-written pronunciation input method of the present invention and known Chinese hand-writing input method.

Chinese character input method of the present invention mainly is to utilize hand-written pronunciation to search the purpose that the Chinese character pronunciation table of comparisons reaches the input Chinese character after identification.Below with an embodiment content of the present invention is described.In following examples, be that the type with palmtop computer (HPC) is that example illustrates, this is because HPC is subject to the cause of volume, generally can't provide very large storage space.Therefore, when the hand-written pronunciation Chinese character input method that uses that present embodiment provided, required storage space can reduce in a large number.Below promptly with regard to the HPC hardware components of present embodiment and realize Chinese character coding input method software section of the present invention, division is as follows.

As shown in Figure 6, it is the block hardware diagram in palmtop computer (HPC) equipment of the embodiment of the invention.Comprise CPU (central processing unit) 20, keyboard 30, external communication interface card 40, LCDs (LCD) 50, lcd driver 60, ROM (ROM (read-only memory)) 70 and RAM (random access memory) 80 among the HPC.ROM and RAM are fixed data (as font storehouse, operating system program etc.) and the temporary data (as user's data) that are used for storing HPC respectively.The user can obtain required information by LCD display curtain 50 when operation HPC, and as input media.And general LCD display curtain 50 to be lcd drivers 60 by special use driven demonstration.In the present embodiment, when CPU 20 carries out hand-written pronunciation input method, obtain importing content by LCD display curtain 50, pronunciation identification system in ROM 70 obtains identification result, be temporary among the RAM 80, again by the pronunciation input method among the ROM 70 according to the pronunciation letter searching Chinese character pronunciation table of comparisons that picks out, just can obtain required Chinese character at last.Can the Chinese character that obtain be presented on the LCD display curtain 50 via lcd driver 60.Based on the hardware configuration of above-mentioned HPC, below be illustrated with regard to the program of actual hand-written pronunciation identification Chinese character coding input method.

Chinese character identification principle figure as shown in Figure 1, (as LCD50 of Fig. 6 etc.) obtains font data by input media 10, becomes standardized binary numeral behind pre-service 111 denoisings.To this binary numeral utilize pattern express 112 extract certain features after, and be stored in standard Chinese word table known in the dictionary 114 and reach form and mate and differentiate 113, just can identify the Chinese words of input.Utilize this principle can make hand-written pronunciation identification system.

Consult the treatment scheme of hand-written pronunciation input method shown in Figure 7, the treatment scheme that realizes hand-written pronunciation input Chinese in the present embodiment is described.As shown in Figure 7, obtain Chinese pronunciation font data (400), after noise is removed in pre-service (401), become standardized binary numeral by input media 10 (as the LCD50 of Fig. 6 etc.) sampling.Again this binary numeral is carried out Filtering Processing (402), extract certain feature (403), directly be stored in standard pronunciation expression-form known in the dictionary database and mate differentiation (404), because Chinese pronunciation letter font is simpler, character set is little, need not production standard radical expression-form and mate differentiation, the identification result that obtains (405) is temporarily deposited among the RAM80.

The Chinese character shown in Figure 4 and the pronunciation table of comparisons are the cores of pronunciation input method, in order to realize the work according to the required Chinese words of pronunciation letter searching of input.The part that this hand-written just pronunciation identification input method is mainly used.

Look into Chinese character shown in Figure 4 and pronunciation to shining according to the pronunciation letter of depositing among the RAM 80, find the identical position (406) of input pronunciation letter in all Chinese character pronunciation tables of comparisons, the Chinese character of this position correspondence is the Chinese character of being asked.So just can draw required whole Chinese character sequences (407).

Again for the example of concrete input " peace " word, be described further below.

Fig. 8 and Fig. 9 represent the example of hand-written pronunciation input method input " peace " word.The phonetic " an " of directly writing " peace " word on lcd screen shown in Figure 8, the identification through hand-written pronunciation identification system obtains Pinyin coding " an ".The ASCII character of " an " is expressed as 01100001b 01101110b, looks into the Chinese character pronunciation table of comparisons shown in Figure 4, can get the Chinese character sequence of " peace " word, as shown in Figure 9, is clicked by the user then.

The present invention is made up known hand-writing input method and pronunciation input method, has created a kind of Chinese character input method of hand-written pronunciation, has following advantage:

Grammatical required storer is more much smaller than the required storer of Chinese handwriting identification character set in the hand-written pronunciation input, and hand-written pronunciation discrimination power is higher, and identification speed is very fast.

Look for Chinese pronunciation button with the use keyboard and compare, can omit keyboard, reduce cost, improve discrimination power, therefore improved work efficiency.

Figure 10 is the contrast table of hand-written pronunciation input method of the present invention and known Chinese hand-writing input method, has shown that the present invention is better than Chinese hand-writing input method before really.

The above embodiment of the present invention is not in order to limiting the present invention, anyly has the knack of present technique person, not breaking away under the spirit and scope of the present invention, can make many changes and modification, and therefore, protection scope of the present invention is as the criterion with appending claims.

Claims

1, a kind of hand-written pronunciation Chinese character input method is characterized in that, comprises the following steps:

A. by the hand-written pronunciation letter of computer related input device input;

B. by this computer the hand-written pronunciation letter of input is sampled and obtain the pronunciation font data;

C. pronunciation font numeral is carried out pre-service, become standardized binary numeral;

D. this binary numeral is made Filtering Processing, and carry out primitive and cut apart, extract the direction and the length characteristic of this binary numeral of expression;

E. directly and be stored in standard pronunciation expression-form coupling known in this computer data and differentiate with the result of steps d;

F. the differentiation result according to the e step searches this computer Chinese character and the pronunciation table of comparisons, finds in the table of comparisons and all identical Chinese characters of input pronunciation;

G. by the computer display unit these Chinese characters are shown for the user and click.

2, the method for claim 1 is characterized in that, this input media and display unit are LCD.

3, a kind of hand-written pronunciation input in Chinese device comprises:

Central processing unit requires 1 described method in order to enforcement of rights;

ROM (read-only memory) is in order to storage identification system, the Chinese character pronunciation table of comparisons, font storehouse, operating system supervisor data;

Random access memory, the temporary data of storage identification result;

Display unit, and by a driver drives, shows the Chinese character of being judged for the user and selects obtaining required information as input media.

4, device as claimed in claim 3 is characterized in that, contains a keyboard, keys in initial consonant and simple or compound vowel of a Chinese syllable for the user, with the input Chinese character.

5, device as claimed in claim 3 is characterized in that, this display unit is LCD.