GB2218838A - Character recognition - Google Patents

Abstract

An information input document or terminal for receiving one or more handformed characters to be input into a data base or processor by a character recognition machine comprises a reference key portion 11 which sets forth a series of characters in a style to be copied by the user and an information receiving portion 12 having one or more pluralities of guide markings 16 arranged to assist the user in forming by hand one or more characters which resemble closely the relevant characters set forth in the reference key portion 11. The characters may be written on a form for later machine reading or may be read as they are written on the terminal. <IMAGE>

Description

IXPROVEMENTS IN AND RELATING TO INFORMATION TRA3FER.

This invention relates to information transfer and in particular, although not exclusively, to a method whereby handwritten characters may be read readily by character recognition means to produce a signal for processing by an electronic data processor, to input means for use in the presentation of handwritten characters to character recognition means and to a processing system for performing the method of the invention.

One of the most common methods of entering information into an electronic data processor involves the provision of a keyboard. This is an efficent method if the actual originator of the information to be entered hes ready access to a keyboard and is able to use the keyboard reliably. Often, however, the originator does not have access to a keyboard or cannot be relied upon to use it properly.

In many situations in which information from the general public or other large group of individuals needs to be entered into an electronic data processor, those people do not have access to a keyboard. Fail order and insurance are examples of business operations in which information commonly is obtained from individuals by sending forms for them to complete in their own homes. Completed forms are returned to a central collection point where a keyboard operator visually reads the forms and enters the information into an electronic data store or processor using a conventional keyboard. This procedure creates the risk of operator errors as well as suffering the disadvantages of being time consuming and expensive.

One procedure for avoiding the use of a keyboard operator involves the provision of optical character recognition means operable automatically to read handwritten characters and generate electronic signals related to the respective characters.

Because of wide variation of handwriting styles it is not possible to use optical character recognition means of the conventional pattern-matching font-dependent type as used for reading typewritten characters of-a pre-selected style. Instead it is necessary to use a much more sophisticated optical character recognition means of the type which is not font-dependent and which relies on a feature analysis procedure to identify a handwritten character by analysis of its geometric features such as the shape and angle aspects of the constituent parts of the character.

The identification of a particular handwritten character by feature analysis requires comparison of its geometric features with information stored in a memory, and identification generally is made on a 'best - fitt basis. There is therefore an undesirable risk of error in identification, and he reading speed is not as fast as that of a font-dependent system. The risk of error may be reduced by a character recognition system which incorporates selfchecking and learning procedures but that leads to a further reduction of reading speed as well as increased costs.

Accordingly it has not hitherto been possible to enter handwritten information into an electronic data store or processor in a manner which is reliable and speedy.

The present invention seeks to provide an improved method for the transfer of information to an electronic data store by the use of a handformed character and means suitable for use in performing at least part of said method.

In accordance with one of its aspects the present invention provides a method for generating an electronic signal related to the identity of a handformed character comprising: -presenting to a user a reference key comprising a series of pre-established characters in a pre-selected style; -providing the user with a character formation surface for the handformation thereon of a selection of any one or more of the pre-established characters, said character formation surface comprising at least one plurality of guide markings to assist the user in creating by hand a selected character in a style which corresponds at least substantially with that character as presented to him by said reference key;; - providing character recognition means, and -operating said character recognition means to generate an electronic signal which is related to the identity of the or each respective hendforneò character.

In a preferred embodiment there is prcvided character recognition means of a kind comprising means to identify 2 handformed character and means to store information concerning one or more of said series of pre-established characters.

The method n2y comprise an acceptance operation whereby the character recognition means or means associated therewith is arranged to give an operator a warning signal if a handformed character cannot be identified as one contained in a reference key series of pre-established characters the details of which are stored in said means . Similarly the method may involve, additionally or alternatively, a verification operation in which 2 warning signal is given if a handformed character, whether for example by virtue of its type or its position on a writing surface, is clearly in error; for example, if a numeral is read in a situation in which a letter should be present.

The method may comprise use of a character formation surface of a kind which is intended to allow the retention thereon of a substantially permanent record of a character handformed on that surfance. Typically a surface of that kinda herein referred to as a character retentive writing surface, may be comprised by a sheet of paper having formed thereon at least one said plurality of guide markings and typically said sheet of paper may also have said character reference key formed thereon. The method may involve use of a writing implement of a kind which deposits a marking medium such as ink, and which normally is visible to the naked eye, or a marking medium which normally is not visible to the naked eye.

In an alternetive the method may comprise use of a character formation surface which is of a re-usable kind not intended to retain a record of a character formed thereon. Typically a surface of that kind, herein referred to as a re-usable writing surface, mEy have a character dram thereon by use of a writing implement which applies local pressure over the writing surface, or which moves 2 light beam over a light sensitive writing surface or which is in the form of monitored position marker, commonly known as a mouse, movable over the writin surface.

Tn a method which utilises a character formation surface in the for of a character retentive uriti - surface the method may frther comprise theprovisionof marking means selectively operable to apply to say a document, or other writing surface which has been presented to character recognition means, a marking which indicates that the document has been read satisfactorily or that there iss say, a verification error as the case may be.The marking preferably is one which makes clear to the originator of the handformed character(s) the nature of the corrective action required. The originator may be provided with a fresh document for completion or the originator or an operator of the character recognition means may be providec with one or more stickers to affix over any character requiring correction.

In this case, preferably the method involves provision of stickers each bearing a pre-formed character though alternatively the stickers may be of a blank kind which bear,for the or each character to be formed, a plurality of said guide markings.

It is envisaged that the method of reading a character retentive writing surface may make use of character recognition means of the optical type. It may be controlled to read and transmit sgnals relating either only to handformed characters appearing on a writing surface or a combination of handformed characters and some or all of any printed characters present on the surface.

The method of the invention may involve the use of an optical character recognition machine of a kind which senses the actual shape of a handformed character as formed with the aid of the guide markings.

In this case it may operate substantially in the same manner as a conventional feature analysis type of optical character recognition machine but because the handformed characters correspond closely with a pre-established range of characters and are in E selected style their identity may be established more quickly and accurately than if free-style handwriting were presented to the machine.

Alternatively the method of the invention may involve the use of an optical character recognition machine of a kind inherently designed, or of the conventional feature analysis type having specially adapter software to rea and interpret only the manner in which the guide markings of a character position have been utilised. The machine or a digital processor connected thereto may then speedily compare information of the utilisation of guide markings with character information held in a memory thereby speedily to establish the identity of a character and facilitate the generation of an electronic signal which is related to the identified character.

In 2 method which utilises a character formation surface in the form of a re-usable writing surface, the method may further comprise the use of character recognition means associated therewith so that there is no need to perform the method step of presenting the writing surface to the character recognition means on each occaision that one or a group of handformed character(s) are to be read by a character recognition means.

Where use is made of a re-usable writing surface and without the use of a writing implement which deposits a visible marking, the method may further comprise providing display means whereby a character being handformed on that surface is displayed during and or following its formation so that the user may verify that the character formed accords with that intended.

In accordance with another aspect of the present invention input means for receiving from a user at least one handformed character comprises: - a reference key portion at which is set-forth a series of pre-established characters in a pre-selected style,and - an information receiving portion at which the user may for at least one handformed character,said infortion receiving portion comprising a character formation surface coTprisingat least one plurality of guide markings to assist thE user in creating by hand a selected character in 2 style which corresponds at least substantially with that character as set forth in the reference key portion.

A preferred form of said plurality of guide markings comprises a plurality of localised markings, such as dots or small squares or circles. Alternatively or additionally other types of guide markings such as elongate markings may be provided. Examples include short rectilinear markings arranged to extend vertically, horizontally and diagonally over the writing surface.

In the case of localised guide markings, these may be arranged as a two dimensional array, a preferred form being 9 dots arranged in 3 equispaced rows each containing 3 equispaced dots for each position at which there is a potential need to form a handwritten character on the writing surface.

In the case of elongate guide markings, preferably these extend vertically, horizontally and diagonally between the neighbouring positions which are occupied by the localised markings of a twodimensional array, or positions which would have been thus occupied had such localised markings been provided.The elongate markings preferably are presented in no more than a semi-bold form whereby an optical character recognition machine is readily able to distinguish between guide markings which have been overlaid in forming a handwritten character and those which have not been overlaid A two-dimensional array of 9 localised markings or elongate markings between 9 such positions is well suited to assisting a user in creating characters if letters of the roman alphabet or arabic numerals are presented in the reference key.Where, however, the reference key contains characters of 2 ncre complicatec form, for example those of the Greek or Russian alphabets, it may be particularly desirable to provide an array of a greater number of markings.

The or each of a group of said plurality of guide markings may be presented within a boundary marking such as a boundary defined by line markings or colour variations thereby to form a box area within which one or more characters may be handforme.

preferably the reference key portion sets forth a series of preestablished characters which comprise the characters of an established alphabet. Those characters may be in a style which results in each alphabet character occupying the fullwidth and height of a plurality of guide markings. Other characters not forming part of an alphabet. eg numerals and punctuation marks, need not necessarily occupy the full width and height of an array of guide markings when handformed; in one embodiment of the invention it is only alphabet characters which occupy the full height and width thereby to enable an optical character recognition machine readily to check that a numeral has not been formed where a letter should have been formed, or vice versa.

ThE seriEs of pre-established characters preferably comprises characters formed in accordance with a set of simple rules thereby to facilitate speedy machine reading of handformed characters that have been formed with the aid of a plurality of guide markings.

OnE suitable set, of rules reflects the above described provision of elongate markings which are arranged to extend horizontally, verticelly and diagonally between neighbouring portions. In accordance with these rules each character is to be formec only by one or more substantially rectilinear portions extending only between neighbouring guide Positions. Thus in a 3 x 3 two dimensional array each character is formed by one or more portions which each follow closely only onE of twenty possible lines of interconnection; it is this limitation in the range of allowable character presentations which greatly assists a machine in speedily reading the handforned characters.

The input means of the present invention may comprise a character formation surface in the form of a character retentive writing surface, such as the surface of a sheet of paper. The writing surface preferably is provided with both said reference key portion and information receiving portion though alternatively the two portions may, for example, be comprised by separate documents.

The input means of the present invention alternatively may comprise a re-usable information receiving portion, referred to as a re-usable writing surface, having formed thereon at least one plurality of guide markings.

It is envisaged that the re-usable surface willbe one which does not require the use of a writing implement that deposits a marking medium on the surface so as to avoid the need for a mechanical erasure operation between the handformation of successive characters.

However, preferably the input means incorporates display means to provide the user with information concerning the character which he hes formed or patt-formed. The display means may present the information remote from or at the re-usable writing surface. In the latter case use may be made, for example, of a character formation surface the localized colouring of which is temporarily changed by the localised application of pressure.That surface may also have associated therewith an array of pressure pads positioned appronriately in relation to the or each said plurality of guide markings so as to be able to detect movement of a pressure pen over the surface and generate an electronic signal related to the identity of the resulting handformed character.

The display means may be a sheet of paper, for example a roll of paper fed over z re-usable writing surface. In that case the paper may be in the form of the aforedescribed character retentive writing surface bearing guide markings and for use with z writino implement which deposits a marking medium; however in this arrangement the paper is not later presented to the character recognition means because use is made of character recognition means associated with the re-usable surface to read characters formed on the overlying paper. Instead of providing guide markings on the overlying paper, that paper may be suitably transparent to allow sight therethrough of the guide markings provided on the re-usable writing surface.

From the foregoing it will be appreciated that the invention envisages use of a range of types of character recognition means. As well as optical pressure-responsive recognition means and position responsive means such as a mouse, use may be made of a grid wire arranged as a pallet digitiser, an ultra-sonic digitiser, a touch sensitive screen utilising for example photo-electric cells, or an infra-red sensing means.

In accordance with yet another of its aspects the present invention comprises an optical character recognition system comprising: -read means for recognition of a handformed character on a writing surface; -memory means for storing information relating to a series of pre-established characters, and -comparison means whereby a signal derived from the read means may be compared with character information stored in said memory means thereby to facilitate generation of an electronic signal related to the identity of the handformed character at said character position.

The memory means may contain information relating to more than one pre-selected style of a series of pre-established characters. Thus it maycontain information relating to two or more alphabets or forms of numerals. In such cases the system preferably comprises character style identification means whereby the system may be set or may set itself to operate in accordance with the information stored in its memory means in respect of a particular character style.

To assist the system in setting itself to read a particular character style the document, form or other input means of the invention may carry a language marking unique to the style of the characters set forth at the reference key portion of the document.

Embodiments of the present invention will now be described, by way of example only, with reference to accompanying diagrammatic drawings in which: Figure 1 is a plan view of an input document in accordance with the present invention; Figure 2 shows part of figure 1 in more detail; Figures 3 & 4 show respectively pluralities of guide markings of input means in accordance with two embodiments of the invention; Figure 5 & 6 correspond to figures 3 & 4 with the addition of handwritten characters; Figure 7 shows a variation of the guide markings of figure 4; Figure 8 is a block diagram of an optical character recognition system in accordance with the present invention, and Figure 9 is a front view of input means in accordance with another embodiment of the present invention.

A simplified single page mail order form10 constituting an input means in accordance with present invention is shown in figure 1 and comprises of an upper reference key portion 11 and a lower information receiving portion 12.

The upper reference key portion 11 displays the 26 letters of the roman alphabet, ten numerals from 0 to 9, and selected punctuation marks. The letters, numerals and marks constitute a series of pre-established characters and each is presented in a particular style.

The style of each character of the series conforms to a set of rules described below.

Each character is presented within a square box defined by a boundary line 13 ( see figure 2 ). Each box contains a two dimensional array of nine uniformly spaced dots 14 and each character is formed from one or more rectilinear portions 15. The style of each character conforms to the rule that the or each said portion 15 of a character shall extend only horizontally, vertically or diagonally to interconnect neighbouring dots. The style thus excludes utilisation of character portions extending, say, from the bottom left dot to the top middle dot. The style also conforms to the rule that each letter and numeral extends for both the full height and width of the array.

The lower information receiving portion 12 comprises in this simplified embodiment only four groups of boxes 16, each group (see also figure 3 ) comprising three boxes. The boxes correspond with the above referred boxes 13 of the reference key portion and each contains a corresponding array of nine dots 14. The nine dots serve as guide means to assist a user in forming by hand, in a box, a character which substantially corresponds in style with a selected one of the characters displayed in the reference key portion 11. Figure 5 shows the boxes of group 16 of figure 3 after insertion of handwritten characters.

Instructions for use of the form 10 and questions to be answered may be printed at either of the portions 11, 12. For mail order use the four groups of three boxes 16, are headed by an instruction to insert the code of required product.

The order form 10 additionally is provided with a signature box 17 for insertion of the user's signature as authority for the order, and the customer's code reference is pre-printed in box 18.

The form 10 additionally displays a language marking 19 relating to the style of the characters displayed by the reference key portion 11.

In the use of the completed form to input information into a digital processor, use is made of an optical character recognition machine of the feature analysis type which is arranged to scan first the language marking 19 and then only the box groups 16 and the customer code reference 18.

As an alternative to dots, each box may contain guide means in the form of an array of short vertical and horizontal lines 20 as shown in figure 4.The lines should be printed in a semi-bold form so that a handwritten character appears distinct. Figure 6 shows these boxes after insertion of the handwritten characters.

Figure 7 shows a variation of figure 4 in which the guide means of each box additionally bears diagonally extending guide lines 20 a.

An optical character recognition system 21 for receiving information for example from the order form 10 is shown in figure 8.

The system 21 comprises optical read means 22, a memory 23 which contains information relating to two series of pre-established characters and a comparison means 24 of the feature analysis type.

The comparison means 24 compares signals derived from the read mesns with information held by the memory means and generates for transmission to a digital processor 25 a series of electronic signals related to the identity of the characters read by the read means.

The read means incorporates a processor control means whereby it reads only prescibed regions of the form 10, and prior to reading any of the groups of boxes 16 of handwritten characters it reads the language marking 19 so that the system operates to recognise the correct style of characters.

In figure 9 there is shown a simplified input terminal 30 of a kind having a re-usable writing surface portion 31 for the sequential handformation of a series of characters with the aid of a pressure pen 35 which in use applies localised pressure to the portion 31.

The input terminal has a reference key portion 32 which corresponds with the reference key portion 11 described above in conjunction with figure 1. It also has a display portion 33 which displays up to a prescribed number of characters, typically up to 30, handformed in sequence on the writing surface portion 31.

The writing surface portion 31 displays an array of nine dots 34 arranged to act as guide means to assist a user in forming by hand a character in a style which substantially corresponds with that of a character in the key portion 32. The dots are arranged in the same manner as the dots of the boxes 16 of figure 1.

The writing surface portion 31 is comprised by a thin layer of tear-resistant plastics which lies over an array of nine pressure sensors ( not shown ) each aligned with a respective dot. Each pressure sensor is arranged to be responsive to localised pressure applied at or close to a respective dot on the writing surface and to assist in the provision of an electronic signal for processing with signals from the other sensors to determine the identity of a character formed by the pressure pen.

The input terminal 30 additionally has a start switch 36, an ' enter button 37 for use when the operator confirms that the display shows the intended character, a ' revise ' button 38 for use when the display does not show the intended character and an ' end ' button 39 for use when the operator has finished entering a set of characters.

Claims (30)

CLAIMS.

1. Method for generating an electronic signal related to the identity of a handformed character comprising: - presenting to a user a reference key comprising a series of pre-established characters in a pre-selected style; providing the user with a character formation surface for the handformation thereon of a selection of any one or more of the pre-established characters, said character formation surface comprising at least one plurality of guide markings to assist the user in creating by hand a selected character in a style which corresponds at least substantially with that character as presented to him by said reference key; - providing character recognition means, and - operating said character recognition means to generate an electronic signal which is related to the identity of the or each respective handformed character.

2. Method according to claim 1 wherein a warning signal is given if a handformed character cannot be read and identified as corresponding to one of the characters in said series of pre-established characters.

3. Method according to claim 1 or claim 2 wherein a warning signal is generated if the character recognition means reads a character which shouldnot have been read according to information stored in the character recognition means or means associated therewith.

4. Method according to any one of the preceding claims wherein use is made of character recognition means of a kind comprising means to read a handformed character and means to store or access a store of information concerning one or more characters of said series of pre-established characters.

5. Method according to any one of the preceding claims wherein the character formation surface is a character retentive writing surface on which a character is handformed by the use of a writing implement which deposits a marking medium on said surface.

6. Method according to claim 5 wherein machine marking means is provided and operated to apply a marking to a writing surface which has been read by the character recognition means.

7. Method according to claim 5 or claim 6 wherein a handformed character is amended by applying thereover a sticker which bears a pre-formed character in the style of a character of the reference key.

8. Method according to any one of the preceding claims wherein use is made of an optical character recognition machine.

9. Method according to any one of claims 1 to 5 wherein the character formation surface is a re-usable writing surface.

10. Method according to claim 9 wherein a character is drawn on said re-usable writing surface by means of a writing implement which does not deposit a marking medium on that surface.

11. Method for generating an electronic signal related to the identity of a handformed character substantially as hereinbefore described with reference to the accompanying drawings.

12. Input means for receiving from a user at least one handformed character comprising: -a reference key portion at which is set-forth a series of pre-established characters in a pre-selected style, and - an information receiving portion at which the user may form at least one handformed character, said information receiving portion comprising a character formation surface comprising at least one plurality of guide markings to assist the user in creating by hand a selected character in style which corresponds at least substantially with that character as set forth in the reference key portion.

13. Input means according to claim 12 wherein said plurality of guide markings comprises a plurality of localised markings.

14. Input means according to claim 13 wherein each of said localised markings is a dot.

15. Input means according to claim 12 wherein said plurality of guide markings comprises a plurality of elongate markings.

16. Input means according to claim 15 wherein said plurality of elongate markings extend between positions which would have been occupied by localised marking had such localised markings been provided.

17. Input means according to claim 15 or claim 16 wherein said elongate markings extend horizontally, vertically and or diagonally over the character formation surface.

18. Input means according to any one of claims 13 to 17 wherein the positions of said localised markings or the positions between which elongate markings extend are arranged as a two-dimensional array.

19. Input means according to claim 18 wherein said positions are arranged as a two-dimensional array of 9 positions disposed in 3 rows each containing 3 positions.

20, Input means according to claim 18 or claim 19 wherein saidpositions are substantially equi spaced in the array.

21. Input means according to any one of claims 12 to 20 wherein the or each of a group of said plurality of guide markings is presented within a boundary marking.

22. Input means according to any one of claims 12 to 21 wherein the reference key portion sets forth characters which are each comprised by one or more substantially recti-linear portions extending only between neighbouring guide positions.

23. Input means according to any one of claims 12 to 22 and comprising a document bearing said reference key portion and information receiving portion.

24. Input means according to any one of claims 12 to 22 wherein said information receiving portion is a re-usable writing surface for the sequential formation thereon of a series of handformed characters.

25. Input means according to claim 24 and comprising display means operable to provide information concerning a character formed on said re-usable writing surface.

26. Input means constructed and arranged substantially as hereinbefore described with reference to the accompanying drawings.

27. An optical character recognition system comprising: - read means for recognition of a handformed character on a writing surface; - memory means for storing information relating to a series of pre-established characters, and - comparison means whereby a signal derived from the read means may be compared with character information stored in said memory means thereby to facilitate generation of an electronic signal related to the identity of the handformed character at said character position.

28. An optical character recognition system according to claim 27 wherein said memory means contains information relating to more than one pre-selected style of a series of pre-established characters.

29. An optical character recognition system according to claim 28 wherein said system may be set or may set itself to operate in accordance with the information stored in its memory means in respect of a particular character style.

30. An optical character recognition system constructed and arranged substantially as hereinbefore described with reference to the accompanying drawings.