JPS61147379A - Optical character reader - Google Patents

Abstract

PURPOSE:To simplify an operation to shorten the correction time and to improve the reading speed furthermore by correcting directly reject characters on a display part and recognizing corrected reject characters. CONSTITUTION:Characters on a form 1 are read by a photoelectric converting part 30 and are converted to electric signals and are stored in the memory of a recognizing part 40. Unrecognizable characters are transferred to a refresh memory 13 from the memory and are stored there and are displayed on a display part 19. An operator corrects the reject characters with a light pen 18 by addition or deletion. Thus, electric signals corresponding to the reject characters stored in the refresh memory 13 are corrected and are transferred to the memory of the recognizing part 40.

Description

Detailed Description of the Invention (Industrial Application Field) This invention is an optical system that optically reads characters written or printed on a form, converts them into electrical signals, recognizes the read characters, and displays them. The present invention relates to a digital character reading device, and particularly to a technique for correcting characters that are determined to be unreadable (also referred to as unrecognizable) as a result of recognition.

(Prior art) Conventionally, optical character reading devices (hereinafter referred to as OCR)
, it is essential to correct characters that are deemed unreadable (hereinafter referred to as reject characters). Regarding the correction of this reject character, for example,
There is a description in Publication No. 78. As shown in FIG. 2, the correction method described here displays the read characters in the character display area A on the screen of the display unit 19, and displays a "?" symbol at the location corresponding to the rejected character. let At the same time, this rejected character is displayed in image display area B. The operator looks at the screen and inputs the correct characters from the keyboard 80. The input characters are then displayed in the area corresponding to the "?" symbol in the character display area A.

In particular, the correction method described in the above publication is characterized in that reject characters are displayed as a rejected multi-value character pattern based on a multi-value digital code that is optically read and converted into an electrical signal. There is.

This correction method allows you to correct the rejected characters one by one while reading the form, so it improves reading speed and It is possible to improve the operability of correcting rejected characters.

(Problem to be Solved by the Invention) However, in the case of the above-mentioned OCR which has a function to correct text characters, the correct characters must be input using the keyboard, so especially when inputting kanji, kana-kanji conversion is required. This method requires complicated operations by the operator and requires a considerable amount of time for correction. As a result, the character reading speed is significantly reduced.

Furthermore, instead of using the kana-kanji conversion method, it is conceivable to display several candidate characters in sequence on the display and input the number of the candidate character using a keyboard. However, this modification method requires a great deal of time to search for and display candidate characters. Therefore, the reading speed of characters ends up being significantly reduced.

Therefore, the present invention solves these conventional problems and provides an OCR that eliminates the need to input reject characters by keyboard operation, is easy to operate, reduces correction time, and further improves reading speed. The purpose is to

(Means for Solving the Problems) The present invention is directed to an optical character reader (OCR) that basically has a photoelectric conversion section, a recognition section, an operation section, and a display section. The photoelectric conversion unit optically reads characters on a form, converts them into electrical signals, and outputs the electrical signals. The recognition unit has a memory and a recognition circuit.

The memory stores electrical signals from the photoelectric conversion section, and the recognition circuit recognizes characters from the contents stored in the memory. The operation section has a refresh memory and means for displaying the contents stored in the refresh memory on the display section.

The refresh memory corresponds to the character code obtained as a result of character recognition, the reject code that should be displayed when the character is unrecognized (rejected), and the character read from the memory and determined to be unrecognizable (reject character). Store electrical signals.

According to the present invention, the following components are provided for the OCR equipped with these components. As a first means, there is provided means for controlling transfer of 1 & 11 data between the memory and the refresh memory. As a second means, there is provided a means for directly adding or deleting the characters displayed on the display section or for correcting them using a light pen. As a third means, the reflation according to the modified content.

Means are provided for modifying the electrical signal stored in the memory.

(Operation) Characters on the form are read by the photoelectric conversion section, converted into electrical signals, and stored in the memory of the recognition section. The contents of the memory are recognized by a recognition circuit character by character. As a result of six character recognition, the recognized characters are converted into character codes and transferred to the refresh memory of the operation unit and stored.

For characters determined to be unrecognizable, a previously prepared reject code is transferred to the refresh memory and stored therein. In this case, at the same time, an electrical signal corresponding to the character determined to be unrecognizable is transferred from the memory via the first means to the refresh memory and stored therein. These pieces of information stored in the refresh memory are displayed on the display section via the means provided in the operation section. As a result, the display shows the positions of recognized characters and unrecognized characters, and a symbol (for example, "?") corresponding to the eject code.
symbols) and characters that are considered unrecognizable (reject characters) are displayed.

Here, the operator uses the second means to correct the reject characters displayed on the display unit by adding or deleting them. According to the revised content, the third
The means for modifying the electrical signal corresponding to the reject character stored in the refresh memory. The modified electrical signal, ie the modified electrical signal corresponding to the ject character, is transferred to the memory of the recognition unit via said first means. This electrical signal is recognized by a recognition circuit. If recognition is possible, the corresponding character code is obtained, transferred and stored in refresh memory. and,'? The character corresponding to the character code is displayed at the position on the display section where " is displayed.

(Example) Hereinafter, the present invention will be described in detail based on an example with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention. In the figure, l is a form, and characters for reading are printed or written on the surface of the form. 2 is the light source, form 1
It is placed in a position to irradiate light onto the top. The light emitted from the light source 2 is reflected on the form l and is input to the image sensor 3 of the photoelectric conversion section 30. The image sensor 3 converts the optical signal into an analog electrical signal and sends it to the A/D conversion circuit 4. The A/D conversion circuit 4 converts the analog electrical signal into a multivalued digital code and sends it to the recognition section 40.

Reference numeral 5 denotes a multilevel memory for temporarily storing a multilevel digital code. That is, the multilevel memory 5 stores a multilevel character pattern string corresponding to one line of characters written on the form l. Reference numeral 6 denotes a binarization circuit, which has a function of binarizing a multi-valued character pattern string at a certain slice level.

A binarization memory 7 stores a character pattern string converted into a z value. A recognition circuit 8 recognizes the binary character pattern string stored in the binary memory 7, converts the recognition result into a character code via the interface control circuit 1O, and sends it to the operation unit 50. Also recognized result reject,
(unrecognizable), a reject code (for example, a "?" symbol code) is sent via the interface control circuit 10, and at the same time, the 2 of the reject character is sent from the binarization memory 7 via the interface control circuit 9. The digitized character pattern string is sent to the operation unit 50. 11 and 12 are interface control circuits for inputting and outputting character codes and binary character pattern strings, respectively.

Reference numeral 13 denotes a refresh memory, which stores the character code (including the reject code) sent from the recognition unit 40 and the binary character pattern string corresponding to the reject character at a predetermined position. The binary character pattern string is stored in the refresh memory 13. It is also possible to perform reverse transfer to the binarization memory 7 via the interface control circuits 12 and 9.

The character codes (including reject codes) stored in the refresh memory 13 are processed by the character generator 14.
, is converted into a display video signal via the switching circuit 15 and the video circuit 10, and is displayed in the character display area A of the display unit 19 shown in FIG. However, if "?" is included in the displayed characters, a reject character is present, so the binary character pattern string of the reject character stored in a predetermined location of the refresh memory 13 is read out, and the switching circuit 1
5 and the video circuit 1B, a reject character is displayed as an image pattern in the image display area B of the display unit 19 shown in FIG.

A light pen 18 has a light-receiving element such as a light-receiving transistor embedded in its tip, detects the scanning line of the cathode ray tube of the display section 19, and sends a detection signal to the light pen control circuit 17. The light pen control circuit 17 has a function of inputting pulses synchronized with screen scanning from the video circuit IB and knowing from which position on the screen of the display section 13 the detection signal from the light pen 18 comes. Furthermore, when the light pen 18 points inside the image display area B, the light pen control circuit 17 writes data of "0" or "1" into the storage area in the refresh memory 13 corresponding to the position of the image display area B. Has a function. Furthermore, when the light pen 18 points to each mode display portion of the mode display area C, the light pen control circuit 17 operates according to that mode.

In this embodiment, there are three modes: addition, deletion, and termination.

Next, the operation will be explained. First, the character reading operation will be explained. In FIG. 1, the irradiated light from the light source 2 is reflected according to the degree of black and white of the form l, and the reflected light is received by the image sensor 3. The reflected light is converted into an analog electrical signal by the image sensor 3, The signal is sent to the D conversion circuit 4. The analog electrical signal is converted into a multivalued digital code by the A/D conversion circuit 4. This multilevel digital code is stored in the multilevel memory 5.

The form I is conveyed in the direction of the arrow H in the figure, and by repeating this process, the multi-value character pattern string corresponding to one line of characters is stored in the multi-value character memory 5. This multivalued character pattern string is converted into a binary character pattern string by the binarization circuit 6. The binarized character pattern string is stored in the z-valued memory 7. The binary character pattern string stored in the binary memory 7 is cut out into a binary character pattern string of one character by the recognition circuit 8.
The digitized character pattern string is recognized. The recognition result is converted into a character code via the interface control circuit 10 and sent to the operation unit 50. In addition, if the recognition result is a rejection, the interface control circuit lO
A reject code (a code with a “?” symbol) is sent to the operation unit 50 through the
The digitized character pattern string is sent to the operation unit 50.

The character code (including the reject code) sent via the interface control circuit 10 of the recognition section 40 is stored in a predetermined location of the refresh memory 13 via the interface control circuit 11 of the operation section 50. Also,
The binary character pattern string corresponding to the object character sent via the interface control circuit 9 is sent to the refresh memory 13 via the interface control circuit 12.
is stored in a predetermined position N. The character codes (including reject codes) stored in the refresh memory 13 are:
It is displayed in the character display area A shown in FIG. 3 on the display section 111I via the character generator 14, switching circuit 15 and video circuit 1B as shown. Furthermore, the binary character pattern strings stored in the refresh memory 13 and corresponding to the object characters are displayed in the image display area B shown in FIG. will be displayed. Note that the display in the mode display area C is performed based on information stored in advance in a storage means such as a ROM (not shown in FIG. 1).

As a result of reading form l by the above character reading operation,
Assuming that the display shown in FIG. 3 has been obtained, the next operation for correcting rejected characters will be described.

In Figure 3, from the character string and "?" symbol displayed in single character display area A, and the character pattern displayed in image display area B, it is easy to see that the rejected character is the character for "ku". It can be understood. It is also conceivable that the cause of rejection is broken lines in characters.

Therefore, using the light pen 18, the light pen 18 is applied to the page position of the mode display section C on the display screen of the 1 display section 18. At this point, the next light pen operation will be in image pattern addition mode.

Next, the light pen IB is applied to the broken part of the character B in the image display area, and the pen is moved so that it becomes the character "ku". The light pen control circuit 17 writes "l tt" data into the refresh memory 13 corresponding to the area on the image display area B where the light pen 18 is applied, and at the same time an image display is added to the area where the light pen 18 is applied. In addition, if dirt or the like on the form 1 is read and rejected, the light pen control circuit 17 is set to the deletion mode by applying the light pen 18 to the chain part in the mode display area C.Light pen By applying the light pen 18 to the dirty part in the image display area B, the control circuit 17 writes data of °"θ in the refresh memory 13 corresponding to the image display, The image display pattern of is deleted from the screen of the display unit 19. The binary character pattern string of the original characters in the refresh memory 13 has line breaks, stains, etc. corrected by the image pattern addition or deletion operation, and has become a normal character binary character pattern string.

Next, move the light pen 18 to the ! in mode display area C. ■ When the part 2 is applied to the refresh memo January 3, the corrected and normalized binary character pattern string of the object character is transferred from the refresh memory 13 to the z-value memory 7 via the interface control circuits 12 and 9. will be forwarded to.

This transferred binary character pattern string was previously stored (in other words, it was obtained as a result of reading from form l)
The rejected characters are replaced with a binary character pattern string.

The replaced binary character pattern string is again sent from the binary memory 7 to the recognition circuit 8, where character recognition is performed. As a result of the recognition, if it is correctly recognized as "ku", the character code of the character "kupa" is transferred to the operation unit 50 instead of the "?" , "?" appears in the character display area A of the display section 19. ” will be displayed instead of ``ku''. If the recognition result is rejected again, the ``?'' code will be displayed in the character display area A, and the code will be modified or the binary character pattern string of the rejected character will be displayed. image is displayed in image display area B. Therefore, the above-mentioned modifications are made again.

The present invention has been described above based on one embodiment.

In the above embodiment, the operation of adding or deleting a part of the image display of rejected characters has been described. However, it is also possible to delete all images of rejected characters and write new rejected characters in image display area B. This is especially effective when rejected characters have lines, stains, etc. that cannot be corrected. It is.

(Effects of the Invention) As explained above, according to the present invention, a reject character can be corrected directly on the display section instead of a keyboard operation, and a reject character can be corrected or recognized as a character. It is possible to provide an optical character reading device (OCR) that is easy to read, can shorten correction time, and has further improved reading speed. This invention is particularly suitable for application to OCR, which performs kanji recognition, and does not require complicated keyboard operations, so it is effective when operated by an inexperienced operator or when a large number of forms are to be processed. be.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a diagram for explaining a method for correcting reject characters in conventional OCR, and FIG. 3 is a display in the embodiment shown in FIG. It is a figure showing an example of a display of a part. DESCRIPTION OF SYMBOLS 1... Form, 2... Light source, 3... Image sensor, 4... A/D converter, 5... Multi-value memory, 6... Binarization circuit, 7...・2nd rectification memory, 8... recognition circuit, 9.1G, 11.12... interface control circuit,
13...Fresh memory, 14...Character generator, 15...Switching circuit, 1B...Video circuit, 17...Light pen control circuit, 18...Light pen, 18...Display section , 30... Photoelectric conversion section. 40... Recognition unit, 50... Operation unit, 60... Keyboard, A... Character display area, B... Image display area, C... Mode display area.

Claims (1)

[Claims] A recognition unit that includes a photoelectric conversion unit that optically reads characters on a form and outputs an electrical signal, a memory that stores the electrical signal, and a recognition circuit that recognizes the content stored in this as a character; a refresh memory for storing a character code read out from the memory, a reject code to be displayed when the character is determined to be unrecognizable, and an electrical signal read from the memory and corresponding to the character determined to be unrecognizable; An optical character reading device comprising: an operation section having means for displaying data on a display section; and a means for controlling data exchange between the memory and the refresh memory; The optical system is characterized by comprising: means for directly adding and displaying or deleting characters with a light pen, and means for modifying the electric signal stored in the refresh memory in accordance with the modified contents. Expression character reader.