JPH0830731A - Method and device for character string confirmation - Google Patents

Abstract

PURPOSE:To shorten the required time of character string confirmation and to improve the accuracy of the character string confirmation. CONSTITUTION:By scanning the preset area of a work surface corresponding to the read rule of a character string and obtaining correlation with the template pictures of preset characters, the position of a leading character is detected and which one of the preset characters is the leading character is confirmed. By scanning a partial area determined based on the relative position of the other character to the position of the leading character and the position of the leading character and obtaining the correlation with the template pictures of the preset characters, the position of the other character is obtained and which one of the preset characters is the other character is confirmed. Until all the characters for constituting the character string are confirmed, scanning within the partial area and the acquisition of the correlation with the template pictures of the preset characters are performed and the acquisition of the position of the other character and the confirmation on which one of the preset characters is the other character are repeated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character string confirmation method and apparatus, and more specifically, it is suitable for confirming a character string (marked character string, etc.) applied to a work for inspection. Character string confirmation method and its device. In this specification, "character string" is used as a term indicating a character composed of hiragana, kanji, letters, numbers, symbols, etc., and "character" means hiragana, kanji, letters, It is used as a term that indicates numbers, symbols, etc.

[0002]

2. Description of the Related Art Hitherto, almost no method has been proposed for confirming a character string applied to a work or the like. By the way, in order to accurately inspect a work or the like, it is necessary to confirm a character string provided by engraving or the like. In order to meet this demand, it is conceivable to adopt a pattern recognition method that has been conventionally applied in various fields. If this method is adopted, since the character string is also a kind of pattern, the confirmation of the character string can be achieved.

[0003]

However, when the pattern recognition method is adopted as described above, not only does the quality of the character string vary considerably among the works, but also the position of the character string considerably. , There is a high possibility that the character string to be confirmed does not match the template image of the character string. Considering such a point, the normal correlation method will be adopted to confirm the character string.
Moreover, since the entire range of the work is set as the search target area, the required time becomes extremely long.

Further, the template image of the character string has a large size as the number of characters increases, so that the time required for calculating the correlation value becomes long, and the time required becomes even longer. Furthermore, if the number of characters forming the character string is large, for example, even if only one character is different from the corresponding image of the template image, a considerably high correlation value is obtained, and erroneous confirmation is likely to occur.

In order to eliminate such inconvenience,
Since a template image must be prepared for each character and the series of processes described above must be performed based on all the template images, the required time becomes even longer.
Also, although it is possible to confirm as a character, it is not possible to confirm as a character string. Of course, it is not possible to confirm a vertically read character string, a horizontally read character string, or a character string extending over a plurality of lines.

[0006]

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and a novel character string confirmation capable of accurately achieving confirmation of a character string and drastically reducing the time required. It is an object to provide a method and an apparatus thereof.

[0007]

According to a character string confirmation method of a first aspect, a predetermined area of a work surface is scanned in accordance with a character string reading rule, and a template image of a preset character is used. By detecting the position of the first character by checking the correlation of the first character, it is confirmed which of the preset characters the first character is, and the relative position of other characters to the preset position of the first character. While scanning the partial area determined based on the position and the position of the first character, and by correlating with the template image of the character set in advance, the position of the other character is obtained and It is confirmed whether the character is a preset character, and scanning and preset text in the partial area until all the characters that make up the character string are confirmed. Performs correlation acquisition of the template image, the position acquisition and other characters of the other character is a method for repeating the confirmation of whether any of the characters that are set in advance.

According to a second aspect of the present invention, there is provided a character string confirmation device which includes a template image holding means for holding a template image of a preset character and a relative position holding means for holding a relative position of another character with respect to the position of the leading character. , Correlating the scanning result obtained by the first scanning means with the scanning image obtained by the first scanning means and the preset template image of the character according to the reading rule of the character string. Based on the first correlation means, the first character position detection means for detecting the position where the correlation value of the first character with respect to the template image is maximum as the position of the first character, and the detected position of the first character and the relative position. Partial area setting means for setting a partial area to be scanned for another character, second scanning means for scanning the set partial area, and second scanning Second correlating means for correlating the scanning result obtained by the step with the preset template image of the character, and the position where the correlation value of the other character with respect to the template image is the maximum, as the position of the other character. Second character position detecting means for detecting, character confirming means for confirming that the template image having the maximum correlation value between the first character and other characters is the first character or another character, and all of the character strings constituting the character string. Until the character is confirmed, it has the partial area setting means, the second scanning means, the second correlating means, the second character position detecting means, and the control means for repeatedly operating the character confirming means.

According to the character string confirmation method of claim 3, the preset template image of the character is divided by the attribute of the character, and the correlation for character confirmation is obtained by
This is a method of correlating with the template image of the character belonging to the corresponding section in response to the attribute of the character to be confirmed being preset. The character string confirmation method of claim 4 is
In order to confirm the character, the correlation value between the character to be confirmed and the template images of all the corresponding characters is calculated, and the template images are ranked with respect to the character to be confirmed.

According to a fifth aspect of the character string confirmation method, template images are sequentially selected according to the ranking, and in the state where each template image is selected, a small area having the same size as the template image is provided in a plurality of directions by a predetermined amount. Calculate the correlation value with the template image by shifting, determine the presence or absence of a small area having a higher correlation value than the small area before the shift, the correlation value higher than the correlation value of the small area before the shift If there is a small area having the same, the above-mentioned correlation value is again calculated and the presence / absence of the small area is determined with reference to the corresponding small area. Determines whether or not the correlation value of the small area before the shift is larger than a predetermined threshold value. If the correlation value of the small area before the shift is smaller than the predetermined threshold value, Te If a plate image is selected and the above series of processing is performed, and the correlation value of the small area before shifting is larger than a predetermined threshold value, the corresponding small area is adopted as the small area of the character to be confirmed. Is the way.

According to the character string confirmation method of claim 6, the inclination of the character string is calculated in response to the positions of at least two characters forming the character string being obtained, and the inclination of the character string is calculated with respect to the position of the first character. This is a method of correcting the relative position with the calculated inclination.

[0012]

According to the character string confirmation method of claim 1, since the scanning is performed in accordance with the reading rules of the character string, the search range for confirming the first character can be considerably narrowed. Since the relative position is preset, the search range can be narrowed sufficiently, and as a result,
The time required for checking the character string can be greatly reduced. In addition, although the confirmation is performed for each character, since the relative position is set in advance as described above, it is possible to accurately confirm the vertical reading, horizontal reading character strings, character strings spanning multiple rows, columns, and the like. be able to.

According to the character string confirmation apparatus of claim 2, since the scanning is performed in accordance with the reading rules of the character string, the search range for confirming the first character can be considerably narrowed, and other characters can be searched. Since the relative position is preset, the search range can be sufficiently narrowed, and as a result, the time required for character string confirmation can be significantly shortened. Also, although confirmation is performed for each character, since the relative position is preset as described above,
You can accurately check vertical reading, horizontal reading character strings, character strings across multiple rows, and columns.

According to the character string confirmation method of claim 3, the preset template image of the character is divided according to the attribute of the character, and when the correlation for character confirmation is taken, the confirmation target character In response to the attribute being preset, the number of template images that should be correlated for each character can be reduced because it correlates with the template image of the character belonging to the corresponding section. The time required for character string confirmation can be further shortened.

According to the character string confirmation method of claim 4, in order to confirm the character, the correlation value between the character to be confirmed and the template images of all the corresponding characters is calculated and the character is confirmed. Since the template images are ranked with respect to the characters, the confirmation process based on the template image with high probability can be preferentially performed, and the time required for confirming the character string can be further shortened.

According to the character string confirmation method of claim 5, for each template image in which the template images are sequentially selected according to the ranking, a small area having the same size as the template image is shifted in a plurality of directions by a predetermined amount. A small area having the highest correlation value with the template image can be obtained, and finally, the small area having the highest correlation value and larger than a predetermined threshold can be adopted as the small area of the character to be confirmed. Therefore, the accuracy of character string confirmation can be improved, and the time required for character string confirmation can be shortened.

According to the character string confirmation method of the sixth aspect, the inclination of the character string is calculated in response to the position of at least two characters forming the character string being obtained, and the other inclination is calculated with respect to the position of the first character. Since the relative position of the character is corrected with the calculated inclination, the area for confirming other characters can be set relatively accurately, and the time required for confirming the character string can be shortened. In addition, it is possible to prevent erroneous confirmation.

[0018]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below with reference to the accompanying drawings showing embodiments. FIG. 1 is a flow chart for explaining an embodiment of the character string confirmation method of the present invention. In step SP1, the leading character of the character string is confirmed by taking a normal correlation with the template image of the preset character, and the position of the leading character is determined. In step SP2, the position of the leading character and the leading character The position of the next character is calculated based on the relative position of the other character with respect to the character, and in step SP3, the next position is calculated by taking a normal correlation with the template image of the character set in advance based on the calculated position. If the next character exists, the position of the area where the normal correlation value becomes maximum is adopted as the accurate position in step SP4, and it is determined whether or not the next character exists in step SP5. For that, the process of step SP2 is performed again. On the contrary, if the next character does not exist, the series of processing is terminated.

A more detailed description will be given. Since it is predetermined whether the character string to be checked is vertical reading or horizontal reading, for example, in the case of vertical reading, the entire range of the search target area is divided into upper and lower areas (Fig. 2 (A in 2), the highest normal correlation value is obtained in order from the uppermost divided area, with the preset template image, and this normal correlation value is higher than the predetermined threshold value. If it is large, it will be obtained as the confirmation result of the first character and the corresponding small area (area of the same size as the template image)
As the first character area. Also, in the case of horizontal reading,
The entire range of the search target area is divided into a plurality of areas on the left and right (see B in FIG. 2), and a normal correlation with a preset template image is obtained in order from the leftmost divided area to obtain the maximum normal correlation value. And the normal correlation value is larger than a predetermined threshold value, it is obtained as the confirmation result of the first character and the corresponding small area is obtained as the first character area. It should be noted that the search in the divided area in A in FIG. 2 is preferably performed in order from the left side, and the search in the divided area in B in FIG. 2 is preferably performed in order from the upper side. The first character can be surely confirmed. Further, the above-described search is sequentially performed in units of small areas, and the small areas that are sequentially searched are displaced from each other by a predetermined amount in a predetermined direction.

Further, it is preferable to add a registered character code to a preset template image, and the character confirmation result can be obtained in the registered character mode. Here, for example, it is possible to add the same registered character code to uppercase letters and lowercase letters. Further, it is preferable to add group data indicating a character group to the template image. In this case,
Character groups are preferably separated according to character attributes such as letters and numbers. If this configuration is adopted, for example, when it is known that the first character is always an alphabet, it is only necessary to take the normal correlation based on only the template images belonging to the alphabet group, and the load of confirming the first character is reduced. In addition, it is possible to shorten the time required to confirm the first character.

Further, the relative position (relative coordinate value) of a small area corresponding to another character is preset as shown in FIG. 3 with the small area corresponding to the first character as a reference (A in FIG. Since the small areas of the first character are B, C, and D respectively indicate the small areas of other characters), after the small area corresponding to the first character is obtained as described above, By adding the relative coordinate value to the coordinate value of the corresponding small area, the coordinate value of the small area corresponding to another character can be obtained (see step SP2). However, since the relative position is not always set accurately, the coordinate values are not necessarily obtained accurately even in the small areas corresponding to other characters. Therefore, with respect to each of the small areas located within the predetermined range (partial area) with reference to the small area obtained based on the relative position, the normal correlation with each template image is obtained, and the normal correlation value is the maximum and the normal correlation is The character can be confirmed based on the template image when the value is larger than the predetermined threshold value (see step SP3). Then, if the characters are confirmed, it is known that this small area is an accurate small area, and therefore the coordinate value of this small area is obtained as an accurate position (see step SP4).

For the second and subsequent characters, if the character group is predetermined, it is only necessary to take the normal correlation based on the template image belonging to the corresponding character group. In addition to being able to reduce the time, it is possible to shorten the time required for character confirmation. Therefore, the time required for character string confirmation can be significantly reduced, and the accuracy of character string confirmation can be improved.

[0023]

[Embodiment 2] FIG. 4 is a block diagram showing an embodiment of a character string confirmation apparatus of the present invention. This character string confirmation device includes a template image holding unit 1 that holds a template image of a preset character, a relative position holding unit 2 that holds the relative position of another character with respect to the position of the first character, and a character string A first scanning unit 3 that scans a preset area on the surface of the work according to the description rule, and a first scanning unit 3 that correlates the scanning result obtained by the first scanning unit 3 with a preset character template image. The correlation unit 4 and the first character position detection unit 5 that detects the position where the correlation value of the first character with respect to the template image is maximum as the position of the first character.
A partial area setting unit 6 for setting a partial area to be scanned for another character based on the detected position of the first character and the relative position, and a second scanning unit for scanning the set partial area. 7, a second correlation unit 8 that correlates the scanning result obtained by the second scanning unit 7 with the preset template image of the character, and the position where the correlation value of the other characters with respect to the template image is maximum. A second character position detection unit 9 for detecting the position of the other character, and a character confirmation for confirming that the template image having the maximum correlation value of the first character or the other character is the first character or the other character. The part 10 and the partial area setting part 6, the second scanning part 7, the second correlation part 8, the second character position detecting part 9 and the character confirming part 10 until all the characters forming the character string are confirmed. Control to make it repeat And a section 11.

The template image holding unit 1 holds a template image of preset characters and, if necessary, holds a registered character code corresponding to the template image and group data indicating a character group. . The relative position holding unit 2 holds the coordinate value indicating the relative position of the small area of the other character with respect to the small area of the first character (area having the same size as the template image). However, this relative position does not necessarily indicate an accurate relative position, but indicates an accurate relative position to some extent.

The first scanning unit 3 scans a preset area on the surface of the work according to the character string description rule. For example, in the case of vertical reading, the entire range of the search target area is scanned. It is divided into a plurality of areas vertically (see A in FIG. 2), and the corresponding area is scanned in order from the uppermost divided area. Further, in the case of horizontal reading, the entire range of the search target area is divided into a plurality of areas on the left and right (see B in FIG. 2), and the corresponding area is scanned in order from the leftmost divided area. Further, in either case, scanning is sequentially performed in units of small areas. The small areas in which the sequential scanning is performed are displaced from each other by a predetermined amount in predetermined directions (vertical and horizontal directions).

The first correlation section 4 obtains a normal correlation between the image obtained by scanning the small area by the first scanning section 3 and the template image. The first character position detector 5
Is for temporarily holding the normal correlation value obtained by the first correlation unit 4 and detecting a small area having the largest normal correlation value and larger than a predetermined threshold value as the small area of the first character. Yes, the coordinate value of a predetermined endpoint of this small area is output as the position of the first character.

The partial area setting unit 6 has a relative position holding unit 2 for the coordinate values output from the first character position detecting unit 5.
The coordinate value of the corresponding partial area is obtained by adding the coordinate values held in the. The second scanning unit 7 sequentially performs scanning in units of small areas within the set partial area. It should be noted that the small areas that are sequentially scanned are displaced from each other by a predetermined amount in a predetermined direction.

The second correlation unit 8 obtains a normal correlation between the image obtained by scanning the small area by the second scanning unit 7 and the template image. However, the second correlator 8 is
The correlating unit 4 may also serve. The second character position detection unit 9 temporarily holds the normal correlation value obtained by the second correlation unit 8, and sets a small area having the maximum normal correlation value and larger than a predetermined threshold as the first character. Is detected as the small area of the character and the coordinate value of a predetermined end point of the small area is output as the position of the corresponding character. However, the second character position detection unit 9 may also function as the first character position detection unit 5.

The character confirmation unit 10 detects and sequentially holds a template image having a maximum normal correlation value and larger than a predetermined threshold value or a corresponding registered character code, and finally stores a series of templates. Output an image or registered character code. Therefore, also in the case of this embodiment, the time required for character string confirmation can be greatly shortened and the accuracy of character string confirmation can be improved.

[0030]

Third Embodiment FIG. 5 is a flow chart for explaining another embodiment of the character string confirmation method of the present invention. Step SP1
In step 1, the leading character of the character string is confirmed and the position of the leading character is determined by taking the normal correlation of the preset character with the template image, and step SP
2, the position of the next character is calculated based on the position of the first character and the relative position of the other character to the first character,
In step SP3, the next character is confirmed by taking the normal correlation with the template image of the character set in advance on the basis of the calculated position, and the position of the area where the normal correlation value is the maximum is accurately determined. And determines in step SP4 whether or not the next character is present. If the next character is present, then in step SP
In 5, the character position correction process is performed on condition that at least two characters have been confirmed, and the process of step SP2 is performed again. On the contrary, if the next character does not exist, the series of processing is terminated.

Further details will be described. The processing of step SP1, the processing of step SP2, and the determination processing of step SP4 are the same as the processing of step SP1, step SP2, and step SP5 in the flowchart of FIG. The processing of step SP3 is the same as the processing of steps SP3 and SP4 in the flowchart of FIG.

The processing of step SP5 is shown by a solid double square and a solid square, for example, even though the relative positions of the second and third characters are set as shown by the dashed squares in FIG. As described above, when the position of the first character and the position of the next character are obtained, the angle θ of the inclination (for example, the inclination with respect to the horizontal direction) of the straight line connecting the position of the first character and the position of the next character is calculated, The calculation shown in Expression 1 is performed to correct the character position as indicated by the wavy square.

[0033]

[Equation 1]

However, x0 and y0 are the coordinate values of the position of the first character, x and y are the coordinate values obtained based on the coordinate values indicating the relative position, and X and Y are the corrected coordinate values. Therefore, in the case of this embodiment, when a name plate or the like is attached to a work by an operator, an industrial robot, or the like, it is suitable when the name plate or the like is inclined, Since the position of the next character is calculated in consideration of the above, the time required to confirm the corresponding character can be shortened as compared with the case where no correction is performed. Of course, when the nameplate is slanted, the amount of correction increases as it moves away from the first character, so the time required to check the entire character string is significantly larger than when no correction is performed. Can be shortened to

[0035]

[Fourth Embodiment] FIG. 7 is a flow chart for explaining a main part of still another embodiment of the character string confirmation method according to the present invention. This flowchart explains the processing when the positions of other characters are obtained with the position of the first character as a reference. A character group is obtained in step SP1, and step SP2
In, the normal correlation value between the template image belonging to the obtained character group and the small area at the corresponding position is calculated, and the priority order is given to the template images in the order of the normal correlation value. While sequentially selecting template images in the order of higher, and confirming the corresponding characters by taking the normal correlation with the template image selected based on the calculated position,
The position of the area where the normal correlation value is maximum is adopted as the accurate position, and the series of processing is terminated.

A more detailed description will be given. In step SP1, for example, when an alphabet group is obtained, only template images belonging to the alphabet group are selected in step SP2 and a normal correlation value with the corresponding small area is calculated. Then, priorities are given to the template images in descending order of the normal correlation value. Therefore, the highest priority is given to the template image having a high probability of being the confirmation target character with a fairly high probability.

Then, in step SP3, a normal correlation value is obtained with respect to a small area at a position displaced by one pixel, for example, from the small area for which the normal correlation value is obtained, and the original normal correlation value is obtained. It is determined whether or not a higher correlation value is obtained. Then, when there is a small area for which a normal correlation value higher than the original normal correlation value exists, the above processing is repeated again with this small area as a reference. By performing this iterative process, a small area that gives the highest normal correlation value to the corresponding template image is finally obtained. Therefore, it is determined whether or not this highest normal correlation value is higher than a predetermined threshold value, and if it is higher than the predetermined threshold value, the character to be confirmed is the same character as this template image. To determine. On the contrary, when it is determined that the highest normal correlation value is less than or equal to the predetermined threshold value, the template image having the next priority is selected and the series of processes described above is performed.

Here, since the small area for which the normal correlation value is obtained to give the priority is located near the true small area of the character to be confirmed, the above-mentioned priority is highly reliable. Since the characters are confirmed according to this priority order, the time required for the character confirmation can be shortened and the accuracy of the character confirmation can be kept high.

[0039]

As described above, according to the invention of claim 1, since the scanning is performed according to the reading rule of the character string, the search range for confirming the first character can be considerably narrowed,
Since the relative positions of other characters are preset, the search range can be narrowed sufficiently, and as a result,
The time required to check the character string can be greatly shortened, and in checking each character, the relative position is preset as described above.
It has a unique effect that it is possible to accurately confirm a horizontally read character string, a plurality of lines, a character string extending over columns, and the like.

According to the second aspect of the present invention, since the scanning is performed according to the reading rules of the character string, the search range for confirming the first character can be considerably narrowed, and the relative positions of other characters are preset. Due to this, the search range can be narrowed sufficiently, and as a result, the time required for character string confirmation can be greatly shortened, and the confirmation is performed for each character. As described above, since the relative position is set in advance, there is a unique effect that a vertically read character string, a horizontally read character string, a character string extending over a plurality of rows, a character string extending over columns, and the like can be accurately confirmed.

The invention of claim 3 has a unique effect that the number of template images to be correlated for each character can be reduced and the time required for character string confirmation can be further shortened. . The invention of claim 4 has a unique effect that the confirmation processing based on the template image with high probability can be preferentially performed, and the time required for confirming the character string can be further shortened.

According to the fifth aspect of the present invention, for each template image in which the template image is sequentially selected according to the ranking, a small area having the same size as the template image is shifted by a predetermined amount in a plurality of directions to form a template image. Obtain a small area with the largest correlation value, finally, the highest correlation value, it is possible to adopt a small area larger than a predetermined threshold as the small area of the character to be confirmed,
There is a unique effect that the accuracy of character string confirmation can be improved and the time required for character string confirmation can be shortened.

The invention of claim 6 has a unique effect that the area for confirming other characters can be set relatively accurately, and the time required for confirming the character string can be shortened. Play.

[Brief description of drawings]

FIG. 1 is a flowchart illustrating an embodiment of a character string confirmation method of the present invention.

FIG. 2 is a schematic diagram showing a method of dividing a search target region according to whether a character string to be confirmed is vertically read or horizontally read.

FIG. 3 is a schematic diagram showing a state in which a relative position of a small area corresponding to another character is set with reference to a small area corresponding to a leading character.

FIG. 4 is a block diagram showing an embodiment of a character string confirmation device of the present invention.

FIG. 5 is a flowchart illustrating another embodiment of the character string confirmation method of the present invention.

FIG. 6 is a schematic diagram illustrating a character position correction process.

FIG. 7 is a flowchart for explaining a main part of still another embodiment of the character string confirmation method of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Template image holding part 2 Relative position holding part 3 1st scanning part 4 1st correlation part 5 1st character position detection part 6 Partial area setting part 7 2nd scanning part 8 2nd correlation part 9 2nd character position detection part 10 Character confirmation unit 11 Control unit

Claims (6)

[Claims] 1. The position of the first character is detected by scanning a preset area on the surface of the work according to the reading rules of the character string and by correlating with the template image of the preset character. At the same time, confirm which of the preset characters is the first character,
While scanning within a preset partial area determined based on the relative position of another character to the position of the leading character and the position of the leading character, the correlation with the template image of the preset character is obtained. By obtaining the position of the other character, it is confirmed whether the other character is one of the preset characters, and the above partial area is displayed until all the characters forming the character string are confirmed. It is possible to repeat the scanning of the inside and the acquisition of the correlation of the preset character with the template image, the position acquisition of the other character, and the confirmation of whether the other character is one of the preset characters. Characterized character string confirmation method. 2. A template image holding means (1) for holding a template image of a preset character, a relative position holding means (2) for holding a relative position of another character with respect to the position of the leading character, and a character. First scanning means (3) for scanning within a preset area on the surface of the work according to the column reading rule, a scanning result obtained by the first scanning means (3), and a template image of preset characters. And a first character position detecting means (5) for detecting the position where the correlation value of the first character with respect to the template image is maximum as the position of the first character.
A partial area setting means (6) for setting a partial area to be scanned for another character based on the detected position of the first character and the relative position; and a second part for scanning the set partial area. Scanning means (7), second correlating means (8) for correlating the scanning result obtained by the second scanning means (7) with the template image of the preset character, and other characters for the template image Second character position detecting means (9) for detecting the position having the maximum correlation value of the other character as the position of the other character, and the template image having the maximum correlation value of the first character or the other character is the first character or the other character. Character confirmation means (10) for confirming that the character is
Until all the characters forming the character string are confirmed, the partial area setting means (6), the second scanning means (7), the second correlating means (8), the second character position detecting means (9) and the character A character string confirmation device comprising: a control means (11) for repeatedly operating the confirmation means (10). 3. A preset template image of a character is divided according to the attribute of the character, and responds to the fact that the attribute of the confirmation target character is set in advance in obtaining correlation for confirming the character. The character string confirmation method according to claim 1, wherein a correlation is made between the characters belonging to the corresponding section and the template image. 4. In order to confirm a character, a correlation value between the character to be confirmed and the template images of all the corresponding characters is calculated, and the template images are ranked with respect to the character to be confirmed. The character string confirmation method according to claim 1 or 3. 5. A template image is sequentially selected according to the ranking, and in a state where each template image is selected, a small area having the same size as the template image is shifted in a plurality of directions by a predetermined amount and correlated with the template image. The value is calculated, the presence or absence of a small area having a correlation value higher than the correlation value of the small area before the shift is determined, and if there is a small area having a correlation value higher than the correlation value of the small area before the shift, , Calculating the correlation value again and determining the presence or absence of the corresponding small area as a reference, if there is no small area having a higher correlation value than the correlation value of the small area before the shift,
It is determined whether the correlation value of the small area before the shift is larger than a predetermined threshold value. If the correlation value of the small area before the shift is smaller than the predetermined threshold value, the template image of the next rank is determined. When the correlation value of the small area before the shift is larger than a predetermined threshold value, the corresponding small area is adopted as the small area of the character to be confirmed. The character string confirmation method described in 4. 6. The inclination of the character string is calculated in response to the positions of at least two characters forming the character string being obtained, and the relative position of the other character with respect to the position of the first character is calculated by the calculated inclination. Claims 1 and 3 to 5 to be corrected
The character string confirmation method described in any of 1.