JPH0132555B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to an optical character reading device that is capable of reading characters, symbols, etc. whose width is larger than the field of view of a sensor. In a general optical character reading device, as shown in FIG. 1, characters, symbols, etc. (hereinafter simply referred to as characters) 2 on a sheet of paper 1 are passed through an optical system 5 and imaged on an image sensor 6 consisting of a photoelectric conversion element. The analog signal from the image sensor 6 is converted into a binary signal by a binary conversion circuit 7 in correspondence with the white background area of the paper and the black of the characters, and the recognition circuit 8 performs character recognition. . In this case, a one-dimensional sensor or a two-dimensional sensor is used as the image sensor 6, and in the case of a one-dimensional sensor, the optical system 5 or the paper 1 is mechanically moved in the lateral direction 4 of the paper 1 to read the characters 2. In addition, when using a two-dimensional sensor, four
The characters 2 are read by being electrically scanned and the optical system 5 or paper 1 being moved mechanically. Both the one-dimensional sensor and the two-dimensional sensor are electrically scanned in the longitudinal direction 3 of the paper 1 to read the characters 2. However, when reading characters optically as described above, in order to capture the characters two-dimensionally with a one-dimensional sensor, the image sensor 6 must be moved in the lateral direction 4 of the paper at a constant speed relative to the paper 1. There is a need. Further, if the moving speed of the image sensor 6 with respect to the paper 1 is not known, it is necessary to store the black/white pattern over the entire character area, and in this case, the storage capacity must be increased, which is expensive. In addition, even if a two-dimensional sensor is used, if the moving speed relative to the paper changes, as shown in Figure 2a,
This method could only be applied when the entire character 2 to be read falls within the field of view 9 of the sensor. For this reason,
When a character 2 larger than the width of the field of view 9 of the sensor is to be read as shown in FIG. It is necessary to store a black/white pattern, and in such a case, the storage capacity needs to be larger than that when a one-dimensional sensor is used. Furthermore, regarding the idea of using a two-dimensional sensor with a large bit configuration to process large-width characters, (i) large-sized two-dimensional sensors are expensive, and (ii) the time required for identification processing is high. There are disadvantages such as increasing the number of For this reason, conventionally it has been difficult and economically disadvantageous to read characters larger than the field of view of the sensor. Therefore, the present invention eliminates the above-mentioned conventional drawbacks, makes it possible to read and identify even characters larger than the field of view of a two-dimensional sensor, regardless of the paper movement speed, and reduces reading errors due to the relative positional relationship between the sensor and characters. The purpose of the present invention is to provide an optical character reading device, and the configuration of the device of the present invention to achieve this purpose includes a photoelectric conversion element that scans characters, codes, etc. and sends analog electrical signals corresponding to the scans. A two-dimensional sensor section, a binarization circuit that converts the analog electrical signal into a binarized signal corresponding to a character area and a background area, and an identification circuit that recognizes characters based on the binarized signal. In the optical character reading device, when the left end of the character to be read is located at the left end within the field of view of the sensor unit scanned using the binary signal, a left screen detection signal is sent out, and the sensor unit sends a left screen detection signal. a left and right screen detection circuit that sends out a right screen detection signal when the right edge of a character to be read is located at the right edge of the field of view of the camera; and a screen feature code extraction circuit that classifies the screen shape from the screen pattern based on the binarized signal. a left screen feature code storage circuit that stores a plurality of left screen feature codes including a left screen and screens before and after the left screen obtained by the screen feature code extraction circuit in response to the left screen detection signal; a right screen feature code storage circuit that stores a plurality of right screen feature codes including a right screen obtained by the screen feature code extraction circuit in response to the right screen detection signal and screens before and after the right screen;
The present invention is characterized by comprising a processing circuit that outputs a screen feature code of one character from the left screen feature code and the right screen feature code. An embodiment of the apparatus of the present invention will be described in detail below with reference to the drawings. The optical character reading device of the present invention forms an image of the character 2 on an image sensor 6 through an optical system 5, as shown in the principle explanatory diagram of FIG.
The background area of the paper is white and the character area is black. Using the binary signal converted into a digital signal by the binarization circuit 7, the left end of the character 2 to be read is set to the sensor. The white/black pattern when the left end of the field of view 10L is reached and the white/black pattern when the right end of character 2 reaches the right end of the sensor's field of view 10R are extracted, and these two are extracted.
Characters are recognized using the black and white pattern on the screen. For this reason, the character 2 to be read
The horizontal axis is the field of view 10R (or 10L) of the sensor 6
Even if the paper is larger, it can be read, and furthermore, the left screen 11L shown in FIG. 3b and the right screen 11R shown in FIG. 3c can be taken out regardless of the moving speed of the paper 1. On the other hand, when character identification is performed using the left and right screens 11L and 11R, when reading using only one left screen 11L and one right screen 11R, for example, the third
Left side 11L ₀ in Figure b and right side 11R ₀ in Figure 3 c
If this is used, depending on the relative positional relationship between each cell 15 of the photoelectric conversion element in the sensor 6 and the image formation of the character element 16, missing portions 13 and 14 of the character element 16 may appear in the binarized signal. may occur, resulting in poor reading. That is, the binary signal is sent to the sensor 6.
Since it is converted into an electrical signal by a plurality of cells 15 arranged vertically and horizontally provided within the field of view 10L (or 10R), the positional relationship between the character element 16 and the cell 15 is shown in Fig. If the line width of the character element 16 is narrow as the area occupied by the character element within the area of each cell 15 as shown in FIG. is cut off, resulting in poor reading. on the other hand,
Screen 11L+1 immediately after movement shown in Figure 4b above
As shown in Fig. 4d, the area occupied by the character element 16 in the same cell expands and is binarized to black (hereinafter referred to as black cell) as shown in Fig. 4e, so that the line of the character The area can be read without being cut. Therefore, in this embodiment, in order to ensure the reading of the character screen, the electrical operation speed of the photoelectric conversion element is sufficiently faster than the relative movement speed of paper 1 and character 2, so that a large amount of screen data can be obtained. In the left end face detection, at least the left screen _11L0 and the screens 11L-1 before and after it,
Similarly, when detecting the right screen, at least the right screen 11R0 and the screens 11R- ₁ and 11R+1 before and after it are used to obtain a binarized signal with no missing character line areas and identify the character screen. This is what we do. As shown in the block diagram of FIG. 5, the optical character reading device of the present invention converts a black and white analog signal obtained by a photoelectric conversion element of a sensor 6 into a digital binary signal by a binary conversion circuit 7. Using the signal 19, a left and right screen detection circuit 20 detects the left and right ends of the characters and outputs a left screen detection signal 26a and a right screen detection signal 26b. Also, the same 2
Screen feature code extraction circuit 2 using valued signal 19
1, column feature codes 52 are extracted for every two columns of a character screen divided into a plurality of rows and columns, and further screen feature codes 27 covering the entire character screen are extracted one after another using the column feature codes 52. , a plurality of screen feature codes 27 are stored in the left screen feature code memory 22.
The left screen detection signal 26a and the right screen detection signal 26b are stored in the right screen feature code memory 23, respectively.
are stored separately. The left and right screen characteristic codes 27L and 27R stored in the left screen characteristic code memory 22 and the right screen characteristic code memory 23 are the combination processing execution instruction signal 2 of the left and right screen detection circuit 20.
6c, the combination processing circuit 24 synthesizes a plurality of identification result candidates, and from among the identification result candidates, the identification character determination circuit 25 determines the identification result to perform character recognition. Each of the above circuits will be explained in detail below. The left and right screen detection circuit 20 converts analog signals read by a two-dimensional sensor 6, which is a two-dimensional array of photoelectric conversion elements, into a binary signal by a binarization circuit 7, as shown in the block diagram of FIG. This binary signal 19 is input to a column-by-column black cell detection circuit 28 that detects the presence of black cells in each column in the vertical direction among the cells of the photoelectric conversion element. The column-by-column black cell detection circuit 28 outputs a leftmost black cell detection signal 30 when it detects a black cell in the leftmost column, and black cells continuously exist over several lines on the left side excluding the leftmost column. When this is the case, the left side character element detection signal 31 is output. Also, when black cells exist continuously over several lines on the right side excluding the rightmost column, a right side character element detection signal 32 is output, and a rightmost black cell detection signal is output when a black cell in the rightmost column is detected. Outputs 33. Note that the several lines mentioned here are the number of lines corresponding to the minimum width of the characters to be read. The above-mentioned column-by-column black cell detection circuit 28 will be described in more detail with the configuration shown in FIG. 7, and the case where the binarized signal is input in parallel (n bits in the horizontal direction) row by row will be described below. . In addition to the binary signals BW ₁ to _{BW o} corresponding to each column in the vertical direction, the column-by-column black cell detection circuit 28 also receives a line clock CK synchronized with the binary signal for each row (1 to m) in the horizontal direction. _1. Frame clock CK ₂ synchronized with the processing of one screen and clear signal CL ₁ slightly delayed from frame clock CK ₂ are input.
In addition, in order to detect black cells, a 1-bit register 28a and an OR circuit _28b are combined for each column.
The data in 8a is updated. Further, 28c is an output register, and 28d is an AND circuit. In such a configuration, each signal in FIG.
The explanation will be given according to the time chart shown in the figure. When the first row data in the binary signals BW ₁ to BW _o is applied to the OR circuit 28b, the first row data is sent to the register 28 by the line clock CK ₁ .
After that, the logical sum circuit 28b calculates the logical sum with the data stored in the register 28a up to the data of the mth row, and sets the data in the register _28a every line clock CK1. Therefore, when a black set exists in a column, the contents of register 28a become "1". In both end columns, the content of the register 28a is directly set to the corresponding output register 28c by the frame clock _CK2 , so the left end black cell detection signal 30 and the right end black cell detection signal 33 are set to the corresponding output register 28c. is set to Also, the contents of the registers 28a on several lines on both sides excluding both ends are added to the AND circuit 28d on the left and right sides, and the output of this AND circuit 28d is set to the corresponding output register 28c by the frame clock _CK2. Therefore, the left character element detection signal 31 and the right character element detection signal 32 are set in the corresponding output register 28c. Also, after the screen data is set in the output register 28c by the frame clock _CK2 in this way,
Register 28a is cleared by clear signal _CL1 . In the above, the white/black pattern on the sensor and
The relationship with the black cell distribution in each column in the vertical direction can be determined by moving the sensor in the horizontal direction, for example, in the 9th column.
9b, 9c, and 9d sequentially from FIG. 9a. However, in Figures 9 a to d, x ₁ to _{x 4} indicate a white/black pattern based on the relationship between the sensor field of view 10 and character 2, and y ₁ to _{y 4} indicate black cells in each vertical column. The distribution of black cells is shown according to the relationship between a portion 36 where there is no black cell and a portion 37 where the black cell is present. Further, the white/black pattern _x1 in FIG. 9a is a case where the leftmost black cell detection signal LBF30 is "0" and the left side character element detection signal LCF31 is "1".
If the moving speed of the paper or the character 2, which is the object to be observed, is slow, a screen with the same positional relationship can be continuously obtained any number of times. For this reason, the following processing is performed to detect as the left screen the screen that first satisfies the above conditions of LBF being "0" and LCF being "1" for a certain character. If the above conditions are met, the left screen flag LF
is set to "1", and set to "0" in other cases. Further, if the left screen flag of the screen immediately before the current detection process is performed is BLF, the left screen detection signal 26a is output when the complement of BLF and LF are .LF=1. Similarly, on the right screen, the white/black pattern x ₃ in Figure 9c is the right character element detection signal RCF.
32 is "1" and the rightmost black cell detection signal RBF33 is "0", and when these conditions are met,
Set the right screen flag RF to “1”, otherwise “0”
shall be. Furthermore, if the right screen flag of the screen immediately before the detection process is performed is BRF, the right screen detection signal 26b is output when the complement of BRF and RF are .RF=1. Left screen detection signal 26
a is input to the left screen feature code memory 22, and the right screen detection signal 26b is input to the right screen feature code memory 23, instructing them to store the respective screen feature codes. Next, the screen feature code extraction circuit will be explained. The principle of the screen characteristic code extraction circuit 21 is shown in FIG. m bits (m
A case will be explained in which the pattern is composed of an n×m bit white/black pattern in a row). First, the two rows on the left
Extract the column feature code content L ₁ of the white/black pattern with V ₁ and V ₂ as a set viewed in the column direction, and then extract the white/black pattern with V ₂ and V ₃ as a set of two columns shifted one row to the right. Extract the column feature code content L ₂ of the black pattern, and then similarly extract the (n-1) set of column feature code content L ₁ to L _o-1 up to the rightmost columns V _o-1 and V _o . , these column feature code contents L ₁ to _{L o-1} are arranged in the order of occurrence to obtain the entire screen feature code contents G ₁ . The screen feature code extraction circuit 21 is composed of a column feature code extraction section and a screen feature code extraction section, as shown in the block diagram of FIG. Below are 2
A case where binarized signals for each row are obtained in parallel in the column direction by the digitizing circuit 7 will be described. The column feature code extracting section includes a parallel-to-serial converter 42 for serially reading out the binary signal 19 input in parallel by the binarization circuit 7, and the parallel-to-serial converter 4
Column feature code contents for every 2 columns by the output of 2
Column feature extraction state transition circuit 44 that extracts L ₁ to L _o-1
and a temporary storage circuit 45 capable of storing intermediate state transition addresses for the number of columns, and the column feature extraction state transition circuit 44 and the temporary storage circuit 45 constitute a sequential circuit. Since the above-mentioned parallel-to-serial converter 42 has a parallel binary signal 19 obtained by the binarization circuit 7,
If this signal is used as it is, (n-1) column characteristic circuits will be required, so it is used for simplicity. Further, the serial binary signal 48 obtained from the parallel-to-serial converter 42 is transferred to the column feature extraction state transition circuit 44.
A temporary delay circuit 43 is provided which delays the serial signal by one bit and inputs the same, thereby inputting the binarized signals for two columns of each row of the screen as one set to the column feature extraction transition circuit 44. Also, the parallel-to-serial converter 42
A write operation is performed by a clock _CK1 synchronized with the output of a binary signal, and a read operation is performed by _{a clock CK3 obtained} by dividing this clock CK1 by n. The temporary delay circuit 43 is operated by the clock _CK3 .
Therefore, for example, a set of binary signals of two columns V ₂ and V ₁ is input to the column feature extraction state transition circuit 44,
In the next state, a set of binary signals from columns V ₃ and V ₂ are input by clock CK ₃ , and sequentially input up to columns V _o and V _o-1 . Furthermore, at the next clock CK ₃ , the binarized signal of the leftmost column V ₁ of the next row is output, so the column
The binarized signals of V ₁ and V _o are inputted, followed by the binarized signals of columns V ₂ and V ₁ . The column feature extraction state transition circuit 44 is accessed by the two columns of binary signals 48 and 49 and the corresponding two columns of intermediate state address data 51 stored in the temporary storage circuit 45, and its output 50 is again sent to the temporary storage circuit. 45. Subsequently, the same processing is performed for the column data of other columns V _i and V _i-1 , and when the binarized signals for the same two columns are input to the data lines 48 and 49, the corresponding intermediate state address data 5
1 is also read out to data line 51a to access column feature extraction state transition circuit 44. By performing these processes m times (for m rows), column characteristic codes L ₁ to L _o-1 of the column screen for two columns are obtained after inputting the m-th binarized signal. For example, the content of the intermediate state transition address is the 12th
As shown in the state transition diagram in the figure, one line 2 in the initial state a.
If the 2-bit binary signal inputs for the column are both white, the state remains in the initial state a, and if the 2-bit binary signal inputs are both black, the state transitions to state b, and then the 2-bit binary signal is converted. If both signal inputs are white, the state transitions to state c. Next, if both 2-bit binary signal inputs are black, the state transitions to state d. Next, if both 2-bit binary signal inputs are white, the state transitions to state c. Transition to state e, then transition to state f if both 2-bit binary signal inputs are black, and if the 2-bit binary signal input is the final stage, state f is the next column feature code 52. It is input to the screen feature extraction state transition section of the second stage. The intermediate state addresses such as a to f are stored in the temporary storage circuit 45, and each time the corresponding binary signals for each row and two columns are input, they are sequentially changed to the final column characteristic code contents L ₁ to L _o-1 is obtained. On the other hand, the screen feature extraction section latches with a screen feature extraction state transition circuit 46 that extracts the screen feature of one character while sequentially transitioning according to the column feature code contents L ₁ to L _o-1 obtained by the column feature extraction section. The circuit 47 constitutes a sequential circuit, and extracts the characteristics of the entire screen. The screen feature extraction state transition circuit 46 receives the column feature code contents L ₁ to _{L o-1} sequentially from the temporary storage circuit 45 (n-1) times, and stores the column feature code contents L _{1 .}
~L _o-1 and the contents of latch circuit 47, and the output is latched into latch circuit 47. Therefore, the contents of the latch circuit 47 change sequentially, for example, as shown in FIG. 22 and right screen feature code memory 23. In Fig. 13, in the initial state a', if there is a column feature code input a'-1 for one set of two columns, the state transitions to state b', and then when there is input b'-1, the state changes to c'. If there is only input c'-1 that overlaps after this state c', the state does not transition and becomes the final stage, and state c' is stored as the screen feature code signal 27 in the left side code memory 22 or the right screen code memory 23. Will be written. however,
The screen feature is a code given to a part or the entire shape of a character to be read, and a plurality of codes are given to one character. There are also cases where the same code is given to some of the different characters. In this way, the screen feature code 27 is extracted at the state transitions shown as A, B, C, etc. in FIG. Next, the synthesis processing circuit will be explained. Here, a case will be described in which three screen characteristic codes are used for the left screen 11L and the right screen 11R. The left screen feature code memory 22 is composed of a three-stage shift register, and the output of the screen feature code extraction circuit 21 is input every time the processing of one screen is completed, and the stored data is sequentially shifted. The writing and shift processing of the left screen feature code memory 22 is stopped by the left screen detection signal 26a, and the left screen feature code memory 22 stores the screen feature codes of the left screen and the screens before and after it, and the synthesis process begins. The screen feature code 27 is held until execution. The right screen feature code memory 23 also has a similar configuration and performs similar processing. Synthesis processing circuit 24
is left screen feature code 27L and right screen feature code 2
By combining 7R, identification result candidates are output, and the left screen 11L and right screen 11R
The combining process is performed in accordance with the combining process execution instruction signal 26c that is input when is detected. For example, as shown in Table 1, when the left screen feature code 27L is A and the right screen feature code 27R is B, "5"
, and the right screen characteristic code 27R is C.
If so, it is identified as "9". Further, when the left screen characteristic code 27L is B or C, and the right screen characteristic code 27R is C, it is identified as "3".
Such compositing processing is performed on three pieces of data in each of the left screen and right screen feature code memories 22 and 23, and the three identification result candidates are "3", "5",
It can be obtained from "9", "? (reject)", etc.

[Table] Next, the identification character determination circuit 25 will be explained. The identification character determination circuit 25 determines the identification result of the character 2 scanned by the image sensor from among the three identification result candidates obtained by the synthesis circuit 24, and is operated by the synthesis processing execution signal 26c. For example, if two results among the identification result candidates are "3" and the other result is "reject", the identification result is set as "3". In this way, even if one identification result results in poor reading, it can be recognized based on the other two identification results. Further, the identification character determination circuit 25 may be constituted by, for example, a microcomputer or the like, and the above processing may be designated and executed as a program command. In this way, the optical character reading device of the present invention can read characters whose width is wider than the field of view of the sensor, regardless of the paper moving speed. Next, other embodiments will be described. In the above embodiment, a synthesis process is first performed, and the identification result is determined from among the plurality of identification result candidates obtained by this synthesis process, but in this embodiment, the screen feature code extraction circuit determines the identification result. A process is performed to determine one optimal screen feature code for each of the left screen and right screen from among the screen feature codes of the left screen or right screen that have been obtained, and the obtained optimal left screen and right screen are The identification result is obtained by synthesizing the screen feature codes.
Incidentally, the same parts as in the previous embodiment are given the same numbers, and their explanation will be omitted. As shown in FIG. 14, in the apparatus of this embodiment, the left screen feature code identification and determination circuit 62 selects the optimal one for the left screen from among the plurality of left screen feature codes 27L stored in the left screen feature code memory 22. The left screen feature code 64 is determined, and similarly, the right screen feature code identification and determination circuit 63 selects an optimal right screen feature code from among the plurality of right screen feature codes 27R stored in the right screen feature code memory 23. 65 is determined. Further, the optimal left screen feature code 64 and right screen feature code 65 obtained by the left screen and right screen feature code identification determination circuits 62 and 63 are combined in a left and right screen feature code synthesis circuit 66 to obtain an identification result. It is. The left/right screen feature code identification determination circuit 62,
The optimum characteristic codes for the left screen and right screen in step 63 are determined by comparing them with the characteristic codes for the left screen and right screen for each character that are stored in the computer 70 in advance. The method of determining the optimal screen feature code for the left side and the right side is, for example, a method of determining the optimal screen feature code by omitting defective screen feature codes from the obtained screen feature codes, or a method of determining the optimal screen feature code for the screen feature code. There are a method of prioritizing and determining, a method of determining by combining the above methods, etc. To explain the above method using the patterns shown in Fig. 3 b and c as an example, in the left screen, in 11L ₀ , the screen feature code cannot be obtained correctly due to the missing part 13, and the code "Reject" is obtained as the screen feature code. It will be done. Let this be P. However, 11L-
1 and 11L+1, since a correct pattern is obtained, a code corresponding to the shape of "〓" is obtained as a screen feature code. Do this T. Similarly, on the right screen, the screen feature code for 11R ₀ is code P for "Reject", and 11R-1,
In 11R+1, a code corresponding to the shape of "〓" (this is designated as U) is obtained. In this way, T with P omitted as the optimal left screen feature code is
U is obtained by omitting P as the optimal right screen feature code. In the above method, a priority order is specified in advance for the screen feature code itself. For example, on the left screen, the code corresponding to the pattern 11L-1 is obtained as T, the code corresponding to the pattern 11L+1 is obtained as T', and if other patterns or codes corresponding to "reject" are obtained, Specify the priority order of T>T′……P,
T, which is designated as the highest priority, is determined as the left screen feature code. Similarly, the right screen feature code is determined as U. The above method specifies the priority order for the left and right screens in advance, and if the feature code of the screen with the highest priority is a code corresponding to "reject", the code of the screen with the next priority is set as the optimal screen. This is determined as a feature code. For example, 11
Assuming that the code corresponding to the pattern of L-1 is T,
The code corresponding to the pattern 11L+1 is T′,
Assume that codes of patterns corresponding to 11L ₀ and others are obtained. In this case, set the priority to 11
_L0 >11L-1>11L+1...... is specified, and when the code of _11L0 is P (reject), the code of 11L-1 is determined as the optimal left screen characteristic code for T. The optimum right screen feature code is determined in the same way for the right screen. Various methods can be considered for determining such screen characteristic codes, but it can be easily done by programming a specific circuit with a microcomputer. Next, the left and right screen characteristic code synthesis circuit will be explained. In the left screen characteristic code identification and determination circuit, the codes Q, R, S, T, U, and P are obtained as shown in Table 2, and from among these, the left screen code T that is optimal for the left screen is obtained, On the other hand, if U is obtained as the right screen characteristic code for the right screen, the left and right screen characteristic codes are combined in the left and right screen characteristic code synthesis circuit 65, and an identification result of "2" is obtained.

[Table] In the above embodiment, the case where the analog signals of the two-dimensional photoelectric conversion element are outputted in parallel in the vertical direction (column direction) was explained, but the case where they are outputted in parallel in the horizontal direction (row direction) is explained. Similar processing can also be performed for . Furthermore, even when analog signals are serially output, similar processing is possible by providing a buffer such as a serial-to-parallel converter at the input section of the left and right screen detection circuit 20 and the screen feature code extraction circuit 21. As specifically explained above with reference to the embodiments, the optical character reading device of the present invention does not require a memory capacity covering the entire character area, and even if the width of the character is larger than the field of view of the sensor, it can read the character regardless of the moving speed. It is possible to read the characters, and reading errors due to the relative positional relationship between the sensor and the characters are reduced, which is further advantageous in terms of cost.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an outline of the configuration of a conventional device, FIGS. 2 a and b are diagrams showing the relationship between characters and the sensor field of view, and FIGS. 3 to 14 are diagrams showing one embodiment of the device of the present invention. , FIG. 3a is an explanatory diagram showing how the same sensor reads characters whose width is larger than the sensor field of view, FIG. a to e are explanatory diagrams showing an example of the positional relationship between each cell of the photoelectric conversion element and the character element, FIG. 5 is a block diagram of the device of the present invention, FIG. 6 is a block diagram of the left and right screen detection circuit, and FIG. The figure is a block diagram of the column-by-column black cell detection circuit, Figure 8 is a time chart of each signal of the left and right screen detection circuits, and Figure 9a
~d is an explanatory diagram showing the relationship between the white/black pattern on the sensor and the distribution of black cells in each column, Fig. 10 is an explanatory diagram showing the processing operation of the screen feature code extraction circuit, and Fig. 11 is the screen feature code. A block diagram of the extraction circuit, FIG. 12 is a state transition diagram in the column feature code extraction section, FIG. 13 is a state transition diagram in the screen feature code extraction section, and FIG. 14 is a main part of another embodiment of the device of the present invention. FIG. In the drawing, 11L and 11R are left and right screens, 13 and 14 are missing text elements, 19 is a binary signal, 20 is a left and right screen detection circuit, 21 is a screen feature code extraction circuit, and 22 and 23 are left screens. and right screen feature code memory, 24 is a synthesis processing circuit, 25
26a, 26b are left screen and right screen detection signals, 26c is a synthesis processing execution instruction signal, 27 is a screen feature code, 27L, 27R are left screen and right screen feature codes, 28 is a black cell by column Detection circuit, 28a is a register, 28b is an OR circuit, 28c is an output register, 28d is an AND circuit, 30, 33 are left end and right end black cell detection signals, 31, 32 are left and right side character element detection signals, 42 is a detection circuit. Para-serial converter, 43 is a temporary delay circuit, 44 is a column feature extraction state transition circuit, 45 is a temporary storage circuit, 46 is a screen feature extraction state transition circuit, 4
7 is a latch circuit, 48 is a serial binary signal,
49 is the temporary delay signal of 48, 51 is the intermediate state address code of the column feature code, 51a is the data line, 52 is the column feature code, 54 is the intermediate state address code of the screen feature code, 62 and 63 are the left screen and right screen Feature code identification determination circuit, 64,6
5 is a left and right screen determination code, 66 is a left and right screen characteristic code synthesis circuit, 70 is a computer, a
~f is the intermediate state address of the column feature code, a′~
c′ is the intermediate state address of the screen feature code, n is the number of columns in one screen, m is the number of rows in one screen, L ₁ to _{L o} are the contents of the column feature code for two columns classified by shape, V ₁
~V _o is the feature code content in each column of one screen,
BW ₁ to _{BW o} are binary signals for one column in each row,
LF, RF are left and right screen flags, BLF, BRF
are the left and right screen flags immediately before the detection process, CK ₁ is the line clock synchronized with the binarized signal for each row,
CK ₂ is a frame clock synchronized with the processing of one screen, CK ₃ is a clock obtained by dividing CK ₁ into n parts, and CL ₁ is a
This is the clear signal for CK ₂ .

Claims (1)

[Claims] 1. A photoelectric conversion element that scans characters, codes, etc. and sends out analog electrical signals corresponding thereto. 2.
A dimensional sensor section, a binarization circuit that converts the analog electrical signal into a binary signal corresponding to a character area and a background area, and an identification circuit that recognizes characters based on the binarized signal. In the optical character reading device, when the left edge of the character to be read is located at the left edge within the field of view of the sensor unit scanned using the binary signal, a left screen detection signal is sent and the sensor unit a left and right screen detection circuit that sends out a right screen detection signal when the right edge of a character to be read is located at the right edge of the field of view; and a screen feature code extraction circuit that classifies the screen shape from the screen pattern based on the binarized signal. and a left screen feature code storage circuit that stores a plurality of left screen feature codes including a left screen and screens before and after the left screen obtained by the screen feature code extraction circuit in response to the left screen detection signal; a right screen feature code storage circuit that stores a plurality of right screen feature codes including a right screen obtained by the screen feature code extraction circuit in response to a right screen detection signal and screens before and after the right screen; An optical character reading device comprising a screen feature code and a processing circuit for identifying one character from the plurality of right screen feature codes. 2. A synthesis processing circuit in which the processing circuit obtains a plurality of identification result candidates by synthesizing the plurality of left screen characteristic codes and the plurality of right screen characteristic codes, and a plurality of identification result candidates obtained by the synthesis circuit. 2. The optical character reading device according to claim 1, further comprising an identification character determining circuit for determining one identification result from among the identification result candidates. 3. The processing circuit includes a left screen feature code identification and determination circuit that determines one left screen feature code from among a plurality of left screen feature codes written in the left screen feature code storage circuit; a right screen feature code identification and determination circuit that determines one right screen feature code from among the plurality of right screen feature codes stored in the code storage circuit; 2. The optical character reading device according to claim 1, further comprising a left and right screen feature code synthesis circuit for synthesizing left and right screen feature codes with a screen feature code and outputting one identification result.