JPS63311491A - Optical character reader - Google Patents

Abstract

PURPOSE:To read the whole line even at a skew angle at which the whole line does not enter into a visual field simultaneously, by obtaining an overall line which has synthesized a recognition line stored in a line storage means, based on a result of a digit alignment processing means, and detecting a replacement of a line which enters into the visual field. CONSTITUTION:When line detecting means 31-33 detect the existence of a line, line storage means 34-36 corresponding to its line detecting means 31-33 store a recognition line obtained from a character recognizing means, a digit alignment processing means 37 allows each character of the recognition line stored in the line storage means 34-36 to correspond to each other by the digit alignment processing means 37, and by a line synthesizing means, an overall line is obtained. On the other hand, when a line replacement detecting means 39 detects a replacement of the line, the contents of the line storage means 34-36 are erased. In such a way, even at a skew angle at which the whole line does not enter into a visual field simultaneously, the whole line can be recognized by an optical character reader.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention is capable of scanning characters, symbols, etc. (hereinafter, only characters will be described as a representative, but the same applies to symbols as well) by scanning a document with a hand-held scanner. The present invention relates to an optical character reading device that reads (a).

[Conventional technology]

At supermarkets, department stores, etc., POS (PaintOf
5ales) system is widespread. This P.

Hand-held optical character readers are often used in S systems.

As such a device, the present applicant has filed a patent application No. 110/1986.
No. 83 and Japanese Patent Application No. 62-56293. FIG. 2 shows a typical configuration of a hand-held optical character reading device.

In Fig. 2, 1 is a scanner, and a hand 2 is used to hold a document 3.
It reads the characters stored in the manuscript simply by matching them with the characters. The document 3 is, for example, a sheet of price tag used in a POS system. 4 is a light source, 5 is a lens system, and 6 is an image sensor. 7 is a control/binarization circuit that converts the analog signal that is the output signal of the image sensor 6 into binary signals corresponding to the character area and the background head. send to

9 to 13 are means for recognizing each character in the screen memory 8 and determining the position (X coordinate) of the character within the visual field.

The screen memory 8 stores binarized data of almost the entire field of view of the image sensor 6. FIG. 3 fat shows an explanation of the binarized data of the image sensor 6.

It is an image sensor with horizontal (X) x vertical (Y) pixels of pxq pixels, and captures characters within its field of view.

Characters and symbols are recognized by the character recognition circuit 13, but since the character recognition circuit 13 recognizes one character at a time, it is necessary to retrieve data for one character from the screen memory 8. - The digit extraction circuit 9 extracts data equivalent to m×q pixels, which is the processing capacity of the single character extraction circuit 11, from the screen memory 8 and stores it in the single digit memory 10. -The character extraction circuit 11 extracts data equivalent to mxn pixels, which is the processing capacity of the character identification circuit 13, from the one-digit memory, and -
It is stored in the character memory 12.

In FIG. 3 (al), the negative digit extraction circuit 9 is
The data from 1 to X-m1Y-1 to y-q is taken out from the screen memory 8 and transferred to the -digit memory 10 (Fig.
bl)). - The digit extraction circuit 9 looks at the contents of the single digit memory 10 and selects a range including character images (in this example, from Y-11 to Y-
11+n-1) for n lines are transferred to the one-character memory 12. (FIG. 3(c,)), - When a character is stored in the character memory 12, the character recognition circuit 13 recognizes the character.

Next, data from X-2 to X-m+1 Y=1 to Y=q is taken out from the screen memory 8 and transferred to the negative digit memory 10 (FIG. 3 (bz)). Then, the image in the range including the character image is transferred to the one-character memory 12. Thereafter, in the same way, shift the position from which to take out screen memory 8 in order, and select single digit memory 1.
0, the image including the character image is transferred to the one-character memory 12, and the character recognition circuit 13 processes it to recognize one line.

The method for determining the range to be transferred from the one-digit memory 10 to the one-character memory 12 is shown in FIG. 4. First, horizontal OR is determined for each row of the one-digit memory 10.

Horizontal OR is an operation that focuses on one row in the horizontal direction and sets it to 1 if there is a black pixel in that row, and sets it to 0 if there is no black pixel. Now, if the no output of the sensor is expressed as 1 and the white output is expressed as 0, then the result of horizontal OR is nothing but the result of taking the logical sum of each pixel in one row, so this operation is called horizontal OR.

In the area where there are characters, the result of horizontal OR is black for that area, as shown in the fourth opening (4th opening). -For example, if the horizontal OR becomes black from Y-13, the range to be transferred from the line memory to the character memory is Y-13, including the white above the character.
−11 to n pixels.

Through the above processing, the
Can read letters and symbols. Image sensor 6
The image of the paper 3 is stored in the screen memory 8 by scanning the image, and the process of recognizing each character in the screen memory 8 is performed three times. 14.15.16, the X coordinate at which the one-digit extraction circuit 9 cuts out the screen in the one-digit memory 10 is stored in the identification result buffers #1, #2, and #3 on 14.15.16.

In the character recognition device shown in FIG. 2, the recognition rate is improved by taking a majority vote from the results of repeated recognition of one character. 14 to 18 are means for taking a majority vote of the character recognition results when the screen is captured and recognized three times. First, the X coordinate values of the characters stored in the recognition result buffers #1 to #3 and the recognition results are set in the digit alignment processing unit 1.
The 0-digit alignment processing unit 17 sent to the 0-digit alignment processing unit 17 associates character recognition results that can be determined to be of the same digit based on the X-coordinate value of the characters. The majority decision processing unit 18 takes a majority decision between the matched character recognition results and obtains the final recognition result for that digit. An example of majority voting is shown in FIG. 5, where (a) is a line written in the manuscript.

(bl) (bz) (bs) are the first and second times, respectively.
Represents the recognition results for the third and third recognitions - (
The character "1" is missing in (bl), the character "2" is missing in (b,), and the character "3" is missing in (b,).
By comparing the recognition results for each digit based on the X coordinate within the field of view of the character and then taking a majority vote (b4)
The correct answer is obtained as follows. Note that if you do not use the X coordinate of the characters and simply match them from the beginning of the recognized characters and take a majority vote, you will not get the correct answer as shown in (c).

23 from 19 to 23 in Figure 2 is a means to output the recognition result that has been correctly read for one line on the manuscript only once.As long as line 0 is within the field of view, the characters are recognized repeatedly. The majority decision result is repeatedly output from 18 onwards.

Figure 6 shows the movement of the field of view and the change in the majority decision result when the scanner 1 is moved from top to bottom while applying the line "C1234567890J". At the position (a,),
The recognition results (b,) are all rejected (unrecognizable: represented by ?). Similarly, the recognition results (bt) at the position (a,) are also all rejected. *At the (as) position, Only the character "0" is recognized within the visual field. The majority vote result from (bl) to (bs) is (C7
), where, for the character “0”, (b
, ) are given priority over the (bl) (b8) paste jets. In other words, more weight is placed on character recognition results than on rejects. The scanner is moved further into the field of view, and the recognition result Cba ) (bs ) (ba ) at the position (a4) (as ) (a= ) is as shown in (C2). Similarly, (a,)(am)(
Recognition result (b,) (bs ) (b
The majority vote result for a) is as shown in (C1).

18 sequentially outputs (c,) (c い) (C3), but 19 is a format checker that outputs the majority result obtained from 18 in a predetermined format (for example, a line starting with C). (C must be followed by 10 numbers).

The timer 20 measures the elapsed time from when a majority result is obtained from the timer 18. If a majority result R+ satisfying the predetermined format is obtained, the previous register 21 and comparator 2
2. The output control section 23 operates as follows. First, the comparator 22 compares R1 with the content R1-1 stored in the previous register 21. If the contents of Ri and Ri do not match, the comparator 22 outputs a NEW signal and the output control section 23 outputs R4 as the recognition result RLI□ of that row.

If the contents of R1 and R1-2 match, the comparator 22 will not output a NEW signal and the output control unit 23 will not output R1 (discard it).
After sending -+ to the comparator 22, R1 is stored. The timer 20 measures the time since the majority vote result is obtained from the timer 18, and erases the contents of the previous register 21 after a predetermined fixed time T eLl has elapsed. The state of the previous register immediately after the power is turned on is the erased state. The TCLI is set to be shorter than the time required to change the price tag (for example, 1 second), for example, about 0.6 seconds (note that if the line is within the field of view and the majority result is repeatedly obtained from 18), The repetition period when this happens is, for example, about 0.2 seconds.

The operations 19 to 23 when reading a price tag will be explained using FIGS. 6 and 7. Majority result of (c,) "?'
Since il'i17'il'7'i'7'i"1lQJ does not satisfy the predetermined format, nothing is output from the format checker 9. Move the scanner from top to bottom and check the majority decision (C2). Result “C1234567
When 890J is obtained, it satisfies the predetermined format, so the formant check section 19 outputs this majority result Ri.The comparator 22 compares Rm with the previous register contents Ri, but the power supply Immediately after inputting , the contents of the previous register have been erased, so R, and R
The contents of i −1 do not necessarily match, so the comparator 22
A NEW signal is output from the output control unit 23, and the output control unit 23 outputs rc1234567890J as the line recognition result RLINE. On the other hand, the previous register 2
rc1234567890J is stored in 1. The majority result of the 0th order (C) is ``C1234567890J, but it is sent to the comparator 22 through the format check section as in the case of (C).However, the previous time Since the content Ri of register 21 is rc1234567890j and matches Ri, the comparator no longer outputs N.
The line recognition result RLI□ is output only once for one line that has entered the field of view, such that the EW signal is not output and the output control unit 23 does not output the RLI□.

Chapter 7 describes the operation when moving the scanner to read multiple lines.
This will be explained using figures. It is assumed that the format check unit 19 has registered the format of each line starting with CSN and ¥. First, the price tag of <a> is rc123456789.
As explained earlier, when the scanner hits the line 0J, the line recognition result RLINI! Moving the scanner down to the 0th order outputs rN123456789
When I applied the scanner to the 0J line, I saw "N12" in my field of view.
When the line 34567890J is entered and the recognition result rN1234567890J is obtained from the majority processing unit 18 for the first time, the previous register content R1 is rc1234.
Since it is 567890J, N is output from comparator 22.
The EW signal is output, and rNl 234567890J is output as the row recognition result RLI□. After that, even if rN1234567890J is repeatedly output from the majority processing unit 18, it matches the contents of the previous register, so R
It is not output as LI□. In other words, “N12345
67890J is output only once. Similarly, “￥
123.456. When the row J comes into view, [¥123°456, J] is output as the row recognition result RLI□. Note that in the PoS price tag according to JIS B9551, each line in one price tag has different contents. Therefore, by comparing the previous recognition result R4-3 with the current recognition result R1 using the comparator 22, it is possible to determine whether or not the same line has been read.

While the price tag is being changed from (a) to (b), there are no characters in the visual field.At this time, no output is obtained from the majority decision processing section 18, and the timer 20 indicates that the majority decision processing section 18 The time elapsed since the recognition result was output is measured, and if the condition in which no output is obtained from each character recognition means continues for more than TCLI because the price tag is changed, the contents of the previous register 21 are erased. Therefore, if you switch the price tag to (b) and apply the scanner to the line ``¥123,456.J,'' ¥123,456.J will be output.

In other words, as can be seen from the explanation of 0 or more, the timer 20 has a function of detecting the exchange of manuscripts (price tags), so that even if the lines have the same content, different price tags can be read consecutively.

[Problem that the invention seeks to solve]

FIG. 8 shows the situation when the scanner 1 is tilted (skewed) with respect to the row and applied to the row.

When the skinny angle θ is small and there is a timing when the entire row is within the field of view like (a), (a,) (as) (
The majority result (C) of the recognition result (b+) (bz) (b, ) for (as) shows that the line recognition result is correct.

Figure 9 shows the situation when the skew angle θ is increased.
It is assumed that the skew angle θ is not so large that a single character completely within the visual field cannot be recognized. Due to the large skew angle, there is no timing when the entire row will be in the field of view, but the right end of the row will be in the field of view with (al) and the left end of the row will be in the field of view with (as), so the majority result (C ), the line recognition results are correctly obtained.

The first graph shows the movement of the field of view and the recognition results when the scanner is slowly moved from top to bottom while keeping the skew angle θ large.
As shown in Figure 0, due to the large skew angle θ, there is no timing for the entire row to be within the field of view, and since the scanner is moving slowly, which adjacent recognition result (b
+ ) (bt4+ ) (b++*) (i is from 1 to 8
Even if you take the majority vote result for the whole line (up to an integer up to), it is not possible to obtain the correct recognition result for the entire line.

As can be seen from the explanation of FIG. 10, in the conventional optical character reading device, when the scanner is moved slowly and the line is captured at a large skew angle so that the entire line is not included in the field of view at once, the entire line is captured. The present invention was devised to solve this problem, and it is possible to recognize the entire line of an optical character reading device even at skinny angles where the entire line cannot be seen at once in the field of view. It aims to make it possible.

[Structure of the invention]

FIG. 1 shows an example of the configuration of an optical character reading device using the present invention. The parts numbered 1 to 23 in the figure have the same functions and configurations as the parts numbered the same in the conventional optical character reading device shown in FIG. In addition to the functions of the digit alignment processing unit 17 shown in the figure, it also outputs the value of the X coordinate of the character after digit alignment.The X coordinate of the character after 0 digit alignment is, for example, the value of the Use the average value of 8 to 18 are the recognition results of each character that entered the field of view obtained from the 0 character recognition means, which is a character recognition means that recognizes each character on the screen captured by the image sensor and outputs its position. In the following explanation, this position information will be referred to as a recognized line.

Reference numerals 31, 32, and 33 are line detection means for detecting whether a part of a line exists in each area when a plurality of areas are set in the field of view. Although FIG. 1 shows an example in which there are three row detection means, a configuration in which there are a plurality of row detection means other than three is also possible. 34, 35, and 36 are line storage means that store recognized lines when the corresponding line existence detection means detects a line. Reference numeral 37 denotes a digit alignment processing means for associating characters with each other based on position information in the recognized lines stored in each of the line storage means. 39
is a line synthesis processing means that obtains one line by synthesizing recognized lines in which correspondences between characters have been added by a digit alignment processing means. In the following description, the line obtained by the line synthesis processing means will be referred to as a comprehensive line. Reference numeral 39 denotes a line change detection means, which is a means for detecting that the lines captured by the image sensor 6 have been changed.

[Effect]

When the scanner 1 is applied to a new line, the line storage means 34, 35, .
The contents of 36 have been deleted. The operations of the line detection means, line storage means, digit alignment processing means, and line comprehensive processing means will be explained below using the example shown in FIG.

Now, as the visual field 30 shows in (a), S7, St,
Ss is divided into three areas, and line detection means #1, #2,
Assume that #3 is determined respectively. When reading the line shown in Figure 11(b) while tilting the scanner, the positional relationship between the field of view and the line and the recognized line are (C+) (c,)Cc, respectively.
s) and (d,)(dg)(ds). (e) shows the memory contents of the line storage means and the result of the line synthesis process when the recognized lines (d,) (dg) (ds) are obtained one after another.

Before applying the scanner to a line, the line swap detection means 3
9, the contents of the row storage means #1, #2, #3 are in an erased state. First, the situation when a line enters the field of view S, as shown in (C,), and the recognized line (d+) is obtained is shown in step (2).

Line detection means #3 detects that the line has entered the area S, and the recognized line is stored in line storage means #3.

Contents stored in each row storage means 101.102.1
03 is subjected to digit alignment processing by the digit alignment processing means 37 (104), and since 101.
05 is the same as 103. Then, the characters of the general line obtained in step 105 are sent to the format check section 19. It is now assumed that the format check section has registered that 12 numbers follow the letters rVJ, corresponding to FIG. 11(b). 105 does not match the predetermined format, so nothing is output from the formant check section.

Next, as shown in Fig. 11 (C3), the situation when the field of view S is entered and the recognition line (dt) is obtained is shown in Fig.
Line detecting means #2 detects that line 0 shown in the step enters the area S, and the recognized line is stored in line storing means #2.

At this time, the contents of row storage means #1 and #3 remain the same as in the Φth stage.

Contents stored in each row storage means 106.107.1
08 is subjected to digit alignment processing by the digit alignment processing means 37 (109), in each digit in the line comprehensive processing! ! The identified characters are selected to obtain a composite line 110. The characters in the 110th general line are then sent to the format check section 19. 110 does not match the predetermined format, so nothing is output from the format check section.

Next, as shown in FIG. 11 (C3), the line enters the 81 area of the visual field and the recognized line of (d,) is obtained. The line detection means #1 detects that the line has entered the line, and the recognized line is stored in the line storage means #1. 111.112.
113 is subjected to digit alignment processing by the digit alignment processing means 37 (114).In the line synthesis process, characters recognized in each digit are selected to obtain the overall line 115.The entire 0 line is the overall line of the stage At this time, it is output via the format check section 19 and the output control section 23 (RL+sx in Figure 1). It is possible to obtain recognition results for the entire row without having to enter the entire line at once.

When the scanner is moved further to read the next line, the line swap detection means #3 detects the line swap, and the line storage means #1. Delete the contents of #2 and #3.

In addition, when the entire line enters the field of view at once as in (f) and the recognized line (g) is obtained, the operation is as shown in (h).
In each region S7, S2, and S3, the row detection means #l, #2, and #3 detect that the row has entered the field of view, so
The recognized line in (g) is stored in each of line storage means #1, #2, and #3 (116.117.118). and,
The entire row is obtained as a result 120 of the column alignment processing/row integration processing.

〔Example〕

An embodiment of the line detection means #1, #2, #3 and the line interchange detection means is shown in FIG. To divide. And in each region R+ −Rs,
The zero line detection means, which sets a range for detecting lines as exemplified by Rs, detects that there is a line within this range. (b) is a circuit implementing the row detection means and the row interchange detection means. 121, 122, and 123 are circuits that perform a horizontal OR operation in the ranges of Rl, sRz, and Rs, respectively.

124.125.126 is a black length determination unit that determines whether the length of consecutive black pixels in the horizontal OR result is within a predetermined range. )
If the horizontal OR result corresponds to that character, the black pixels are consecutive for the height of the character, so the black length determination unit determines whether the character is in that area, and The result is output to the EXISTi signal line. EXIST,
The signal f! is sent to the corresponding row storage means, and each row storage means receives f! Gives you the timing to memorize your knowledge. 127 is EX
This is an OR gate that takes the logical sum of TST+, EXISTx, and BXIST3. 127 output EX I ST
becomes true if there is a character in any of R,, Ro, and Rj.

To detect a row swap, it is sufficient to detect that the row leaves the field of view, that is, when BXIST becomes false. 128 inverts the logic of EX I ST and outputs the row swap signal. This is an inverter gate that creates ALL CLR. The ALLCLR signal is sent to all row storage means and provides the timing for erasing the contents of the row storage means.

When the row is not within the field of view, the majority decision processing unit 18
Therefore, an embodiment is also possible in which the fact that no recognized line is obtained is detected and it is determined that the lines have been swapped. FIG. 13 shows an embodiment in which it is determined that a row has been replaced by measuring the time during which a recognized row is not obtained and the time exceeds a predetermined time (referred to as TL). In FIG. 13, the timer 130 If a row is not obtained for a period longer than TL, a T Lup signal is output. TL is the period (e.g., 0.
2 seconds) (for example, 0.25 seconds).

In FIG. 13, the parts numbered 1 to 23 have the same functions and configurations as the parts with the same numbers in FIG. 6
is an image sensor, and as in the case of Figure 2, it requires a field of view of at least one line of characters written on paper 3, and in Figure 13, the width is one line and the height is about three times the length of one character. There is. In addition, if multiple lines are written on the document, complex processing is required if multiple lines enter the field of view at once, so to simplify the processing system, the height of the image sensor's field of view should be set above the document. It is better to make it smaller than the line spacing.

In FIG. 13, the line detection means #1 to #3, the line storage means, the digit alignment processing means, and the line synthesis processing means are implemented using a microprocessor 131, ROM 132, and RAM 133. The ROM 132 stores a program for the microprocessor 131 for implementing the line detection means, digit alignment processing means, and line comprehensive processing means. The zero line storage means 34 is implemented in a variable area set on the RAM 133. .

A schematic flowchart of the processing of the microprocessor 131 is shown in FIG. 14. When the power of the optical character reading device is turned on, the processing starts from ①. (2) is a process for erasing the contents of the row storage area provided in the RAM. ■ is timer 130
In the process of determining whether or not the is outputting a TLUP signal,
TLt+? When the signal is being output, that is, when there is a change in the rows, the process proceeds to step (2), in which it is determined whether or not a recognized row has been obtained. ■Ha! ! This is the process of reading the information. From ■, [stock] has implemented line detection means,
This is a process of dividing the field of view into a plurality of areas, detecting the presence of a line in each area, and storing the recognized line in the line storage area corresponding to each area. The flowchart in FIG. 14 divides the visual field into three regions S1, St, and Ss (
The process of storing the tgai line in the line storage area #1 corresponding to the esI area, which is the processing process in the case of division in Figure (a)), is 0. ■Whether or not the number of characters recognized (not rejected) in the Sl area of the recognized line is greater than the number of characters recognized in the Sl area of the content stored in line storage area #1. This is the process of determining. ■■,
■[Phase] indicates that the same processing as ■■ is performed for the St and Ss regions, respectively. ■ is row storage area #l
This is a digit matching process for finding characters that correspond to each other based on the position information of the recognized characters stored in #3. @ is a process in which comprehensive processing is performed based on the result of digit alignment performed in ■ to obtain a comprehensive row. 0 is a process of sending a comprehensive line to the format check unit 19.

16 to 18 show detailed flowcharts of the process shown in FIG. 14. In the following explanation, variables and constants are used as shown in FIG. 15. That is, as shown in FIG. 15(a), the range of the X coordinate of each region of Sl, St, and Ss is from X to X8, and from X to X.

, X, to X4. Also, as shown in table (b), i! ! Line m has N characters, and the characters and positions are a
+:), xf:l, the contents stored in row storage area #1 include I characters, and the characters and positions are expressed as al, x(
! i-1, 2, ......, ■), also expressed as l,
Similarly, for 0-line storage areas #2 and #3 where the number of recognized (non-rejected) characters in the area of Sl is expressed as Pl, J% a7, Xj-, pg and Ksa*sx':)
The number of digits after zero digit adjustment is expressed as , P, and the position of each digit is y, and the combination of letters is,
Let it be expressed as bH and bl. It is assumed that the number of characters in the total line is M, and the character is represented by C1.

FIG. 16 shows a detailed flowchart of the process shown in FIG. ■■■ is■
This is an iterative process for repeating the process of ■■. (2) is a process for determining whether or not a7 to be processed in (2) falls within the area of Sl. ■■ is a process in which the number of p is increased by 1 when all is not rejected (represented by a ? symbol).

(2) is a process of determining whether p is greater than P.

The processing from ① to ② is the detailed processing of ② in FIG. (2) is a process in which p is registered as a new P+ and N' is registered as the number of characters (2) to be stored in the row storage area #1. @■@0 is a process of storing the recognized line in the line storage area #1.

FIG. 17 shows a detailed flowchart of the digit alignment process shown in FIG. 14.
,■ is used for initialization. (2) This is a process in which the end of the contents stored in the brush storage area is indicated in advance by a special large value X11. Set X14 to a value larger than the sum of X4 in Figure 15(a) and the value of W, which will be described later.■
is the coordinate x of the i-th character in the row storage area #1 that we are looking at
3: ゝ, coordinates of the ``th character in row storage area #2, X<i',
Minimum value Xm1 of the coordinate xT of the second character in row storage area #3
This is the process of finding n. In case (2), if X■in is the same as XN used in the process (2), it is determined that the digit alignment process for all characters has been completed. W used in (2) indicates the width of the X coordinate that can be determined to be the same digit. Now, if xl:'≦X■in+W holds, then xT
performs the processing when it falls into the digit currently being processed, that is, the processing of ■ and ■.■ is registered as the i-th character a (character bt after digit adjustment of 11) in row storage area #1. It is processing. (2) is a process in which the value of i is increased since the first process has been completed.

On the other hand, if x11゜ does not fit into the column currently being processed, process ■ is performed. ,
(# has the same meaning as the space between characters in Fig. 5(b)). In the same way as ■ to ■, register for line storage area #2 and line storage area #3. Processing■〜＠
, 0 to ■.

O is a process of registering Xm1n in y as the position after digit alignment. [Phase] is a process in which the number l of characters after digit alignment is increased by one. [Phase] is a process in which the finally obtained number of characters z-1 is registered as a variable.

FIG. 18 <a> is a detailed flowchart (part 1) of the general processing shown in FIG. ■■@ is an iterative process in which the processes from ■ to ■ are repeated as many times as the number of digits after digit alignment. Based on ■, it is determined whether the coordinate y of the first digit of interest falls within the area of Sl, and S
When entering the area of , proceed to ■, and in ■[F],
If the character bT11 in the line storage area #1 after alignment is not # (if there is no missing character), a process is performed to register it as the character C in the general line. ■So, y.

It is determined whether or not it enters the area of Sl, and if it enters the area of S2, proceed to ■. In ■■, the row storage area #2,
If the character S after alignment is not # (if there is no missing character), a process is performed to register it as the character C1 in the general line. If it reaches [stock], y enters the area of Ss, and in 0■, if the character S after column alignment in row storage area #3 is not # (if no character is missing) , a process is being performed to register it as the character C1 of the general line. (2) is a process of registering the number of characters (m-1) in the general line in the variable M.

FIG. 18(a) shows the process when the row storage areas #1, #2, and #3 used for each area S+, St, and Ss are divided, but it is an example of processing in which the row storage areas used for each area are not limited. is also possible, as shown in FIG. 18(b). In FIG. 18(b), ■ is a process of registering the number of digits after digit alignment as the total number of characters M in the line. ■■[Phase] is the process from ■ to 0.
This is an iterative process that is performed repeatedly.

In ■0■ of Fig. 18(b), b is a missing character (
b(1
1 to C1. When b'', ' is a missing character jet, proceed to ■, and when it is neither a missing character nor a reject, proceed to C6. b,
If is a missing character jet, proceed to [phase], 0■O
In this case, when ``su'' is neither a missing character nor a reject, ``su'' is set to C1. When b is a missing character jet, it advances to O, and it goes to 0 after all.

All of these are rejected or missing characters, and in this case, the reject is set to C5.

Note that the initialization process of ■ in Fig. 14 is actually performed by ■, J
,, 1P1,P! , P3 should be set to 0.

In the above embodiment, the number of areas set in the field of view is 3.
Although the case is shown in which there are a plurality of areas, an embodiment in which areas other than three are set is also possible. Also, when setting multiple areas in the field of view, mutually exclusive (so that they do not overlap)
It is not necessary to set these, and for example, as shown in FIG. 19, an embodiment in which five regions are set while partially overlapping is also possible.

〔Effect of the invention〕

By using the present invention, it is possible to realize an optical character reading device that can read an entire line even at a skew angle that prevents the entire line from entering the visual field at once. Conversely, this means that the conventional optical character reading device (Fig. 20(a))
This means that the height of the field of view can be made smaller compared to (Fig. 20(b)).

Even with a scanner having the field of view height shown in FIG. 20(b), by using the present invention, the entire line can be read by applying the scanner to the line and moving it up and down.

Reducing the height of the field of view has the following effects.

- Since the number of pixels required for the image sensor is reduced, a cheaper image sensor can be used and the cost of the device can be reduced.

- By reducing the field of view height, the size of the scanner becomes smaller and the operability of the scanner increases.

- Because the field of view height is smaller, the illumination light source inside the scanner is less likely to cause unevenness when illuminating the document.
Design and development becomes easier.

As described above, the ripple effects brought about by the present invention are large.

[Brief explanation of the drawing]

FIG. 1 shows an example of the configuration of an optical character reading device using the present invention.
Fig. 2 is an optical character reading device according to the prior art, Fig. 3 is an explanatory diagram of the process up to single character extraction processing, Fig. 4 is an explanatory diagram of the single character extraction method, Fig. 5 is an explanatory diagram of digit alignment/majority result processing, Figures 6, 8, and 9 are illustrations of the movement of lines within the field of view and majority voting results, and Figure 7 is an illustration of price tag reading.
FIG. 10 is an explanatory diagram of the problems of the prior art, FIG. 11 is a detailed explanatory diagram of the present invention, FIG. 12 is a diagram of an embodiment of the line detection means and line interchange detection means, and FIG. 13 is an illustration of the present invention. Example diagram,
Fig. 14 is a schematic flowchart of the microb convex processor processing, Fig. 15 is an explanatory diagram of constants and variables, and Figs. FIG. 19 is an explanatory diagram of how to set the area, and FIG. 20 is an explanatory diagram of the field of view of the optical character reading device using the present invention. 1...Scanner 2...Hand 3.
...Original 4...Light source 5...
... Lens system 6 ... Image sensor 7 ... Control binarization circuit 8 ... Screen memory 9 ... - Digit extraction circuit 10 ... ...
・-Digit memory 11...-Character cutting circuit 12...-Character memory 13...Character recognition circuit 14.15.16...Recognition result buffer 17 .17'... Digit alignment processing section 18...
...Majority processing section 19...Format check section 20.130
... Timer 21 ... Previous register 22 ... Comparator 23 ... Output control section 30 ... Image sensor field of view 3L 32.3
3...Line detection means 34.35.36...
・Line storage means 37...Column alignment means 38...Line comprehensive processing means 39...Line interchange detection means 101.106
．． 111.116... Contents of line storage means #l 102.107.112.117... Contents of line storage means #2 103.108.113.118... Lines Contents of storage means #3 104.109.114.119... Digit alignment result 105.110.115.120... Line total result 121.122.123... Horizontal OR circuit 12
4.125.126... Black length determination section 127.
... OR gate 128 ... Inverter gate 131 ... Microprocessor 132 ...
...ROM 133...RAM. Patent Deseto Sumitomo Electric Industries Co., Ltd. Agent Sickle 1) Text 2 Figure 3 Figure 5 Figure 4 (a) ¥12345 Figure 6 n French 5) Mouth - Mouthful (b,) C1234567890,8
) Figure 7 Figure 8 Figure 9 Figure 10 Figure μ7 Cow Figure 15 (a) Figure 16 Figure 18vA (a)

Claims (7)

[Claims] (1) In an optical character reading device that reads characters on a document by holding an image sensor housing (scanner) in your hand and applying it to the document, an image of how multiple characters, such as one line of characters or symbols, can be captured within the field of view. A sensor, a character recognition means that recognizes each character on the screen captured by the image sensor and outputs its position, and a system that determines whether a portion of a line exists in each area in multiple areas set in the field of view. A line detection means for detecting a line, a plurality of line storage means for storing recognized lines obtained from the character recognition means when the corresponding line detection means detects a line, and a recognized line stored in each of the line storage means. , a digit alignment processing means for matching characters with each other based on character position information, and a line synthesis means for obtaining a comprehensive line that integrates the recognized lines stored in the line storage means based on the result of the digit alignment processing means. and line transposition detection means for detecting a transposition of lines within the field of view, and when the line detection means detects the presence of a line, the line storage means corresponding to the line detection means is obtained from the character recognition means. The digit alignment processing means associates the characters of the recognized line stored in the line storage means with each other, and obtains a comprehensive line by the line synthesis means.Meanwhile, the line interchange detection means 1. An optical character reading device which erases the contents of a line storage means when detecting a line change. (2) In the optical character reading device according to claim 1, the line detection means includes a horizontal OR circuit for calculating the logical sum of black pixels in the horizontal direction in a predetermined range provided in the field of view of the image sensor; It consists of a black length determination section that determines whether the calculation results of the OR circuit are continuous for a predetermined range of length in the vertical direction, and the black length determination section detects the calculation results of the horizontal OR circuit that are continuous for a predetermined range of length. An optical character reading device characterized in that it sometimes determines that a line exists. (3) In the optical character reading device according to claim 1, the line detection means checks the positional information of each character in the recognized line obtained from the character recognition result, and determines whether the character is not unrecognizable and has positional information within a predetermined range. It is a means for determining that a line exists based on the presence of a character, and there are more non-unrecognizable characters in the line than there are non-unrecognizable characters having position information in the range in the recognized line already stored in the line storage means. When the range is obtained,
An optical character reading device characterized in that a newly obtained recognized line is stored in a line storage means. (4) In the optical character reading device according to claims 1 to 3, the line transposition detection means determines that a line transposition occurs when none of the line detection means detects a line. An optical character reading device featuring: (5) In the optical character reading device according to claim 1 or 3, the line change detection means measures the time during which no recognized line is obtained when the character recognition means repeatedly performs recognition processing. What is claimed is: 1. An optical character reading device comprising: a timer for determining that a line has been swapped when a recognized line is not obtained for a predetermined period of time; (6) In the optical character reading device according to claims 1 to 5, the height of the field of view of the image sensor is 1 to 3 times the height of the character.
An optical character reading device characterized in that the positional information of a character used for digit alignment is horizontal positional information within a visual field of the character. (7) In the optical character reading device according to claims 1 to 5, the height of the field of view of the image sensor is smaller than the line spacing between the character lines written on the document, and the position is used for digit alignment. An optical character reading device characterized in that the information is horizontal position information within a field of view of the character.