JP2001043317A - Character inspection device and its dictionary automatic generation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce man-hours at the generation of multi-font dictionary for inspecting a character. SOLUTION: The reference font of a prescribed character is cut from a picture taken in by a video camera 12. A dictionary picture processing part 19 executes enlargement, reduction, rotation and a processing obtained by compounding them on the reference font in accordance with a parameter which is set by a parameter setting part 18. They are stored in a dictionary memory 20 with the reference font. For automatically generating the other font from the reference font, man-hours are reduced.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character inspection apparatus and a dictionary automatic generation method therefor, and more particularly to a method and a method for automatically distorting or missing characters or symbols (hereinafter simply referred to as "characters") written on cardboard or the like. The present invention relates to a character inspection apparatus for inspecting shape determination by comparison with a multi-pattern registered in a dictionary in advance, and a dictionary automatic generation method for registering the dictionary.

[0002]

2. Description of the Related Art In an inspection apparatus for a character shape written on a cardboard or the like, a non-defective sample density image including a character to be inspected is captured by a video camera or the like, and a range to be used as a dictionary is selected from the image and registered in the dictionary. Keep it. When performing the inspection, the character shape is collated by pattern matching between the image to be inspected captured by a video camera or the like and the image representing the character shape (font) of the dictionary, and the degree of coincidence as the collation result is determined to perform the character inspection. Do. In this inspection method using pattern matching, if only one character pattern is prepared for one character in the dictionary, matching cannot be performed well when the character to be inspected is enlarged, reduced, rotated, etc. The matching degree between the character and the dictionary is reduced.

As one method for dealing with such a problem, there is a use of a multi-font dictionary in which a plurality of fonts are prepared for one character in a dictionary, and a matching judgment is made with those fonts. As a known example of such a technique, for example, there is a “character deformation method and character recognition device” disclosed in Japanese Patent Application Laid-Open No. 7-65127, in which a stamp character and a stamp character are deformed due to crushing, chipping, inclination, displacement, and the like. It shows how to create and use character fonts.

[0004]

In order to register a plurality of fonts for one character in the dictionary, a character font is created for each expected deformation state of the character, and an image is input and the image is input. Operations such as the specification of the range to be used as the dictionary and the specification and registration of the font number of the registration destination have to be repeated, which has a problem that it takes time and effort. In this regard, the technique disclosed in Japanese Patent Laid-Open No. 7-65127 described above creates one reference font by performing the following processing for image input of only one font as one reference for one character. Multi-font dictionaries are created by using image processing based on parameters specified by hand to create fonts that are used when crushed, missing, skewed, or misaligned stamp characters occur. Time and labor can be greatly reduced. However, in the case of a stamped character on a cardboard or the like, the size of the imaged character itself may change due to changes in imaging conditions for character inspection rather than crushing or chipping of the character. Does not disclose a multi-font generation method including such enlargement and reduction of characters.

According to the present invention, when a font of one character is generated by one imaging and registered, a font when a deformation including enlargement and reduction of the character is performed is automatically generated and registered to create a multi-font dictionary. It is an object of the present invention to provide a character inspection apparatus capable of generating a dictionary and a method for automatically generating a dictionary.

[0006]

SUMMARY OF THE INVENTION The present invention comprises a dictionary memory in which a plurality of fonts are registered for one character, and stores a pattern of an input inspection target character and each of the fonts registered in the dictionary memory. In a character inspection apparatus configured to perform character inspection by comparing and matching, a parameter for deforming a font of a standard character captured and created by a character image input unit by at least image processing including enlargement and reduction processing is performed. Parameter setting means for setting, and font image processing means for performing image processing including at least enlargement and reduction in accordance with the parameters set by this means, generating a transformed font and registering it in the dictionary memory, A character inspection apparatus characterized by comprising:

Further, according to the present invention, the parameter setting means preferably sets a parameter indicating at least a vertical enlargement ratio, a horizontal enlargement ratio, a vertical reduction ratio, and a horizontal reduction ratio of the standard character. A character inspection apparatus characterized in that the character inspection apparatus is capable of performing the following.

Further, the present invention provides a dictionary automatic generation method for a character inspection apparatus for registering a plurality of fonts of the same character in a dictionary for character inspection, wherein a reference font of the character is extracted from image data fetched by a character image input means. Generate
Further, another character font is generated for the character by performing image processing including at least one of enlargement and reduction processing on the reference font, and registered in a dictionary. A generation method is disclosed.

[0009]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing an example of the configuration of a character inspection apparatus according to the present invention. The video camera 12 captures characters 11a written on the surface of an inspection object 11 such as a cardboard, and the inspection object 11 is a belt. Sensor 1 that detects that it has moved to a predetermined position by a conveyor or the like
3 and an image processing device 23 for determining the character shape of the character 11a and the like.

An image processing device 23 receives a signal from the sensor 13 and instructs the video camera 12 to capture an image.
Video capturing unit 15 for capturing video from the camera, and image memory 16 for storing video captured by video capturing unit 15
A dictionary registration unit 17 for registering a dictionary using images on the image memory 16, an automatic dictionary generation unit 24 for registering fonts subjected to image processing at the time of dictionary registration, and a dictionary memory for registering fonts 20; a matching unit 21 for performing pattern matching between an image registered in the image memory 16 and a font registered in the dictionary memory 20; and a result determining unit 22 for determining a result obtained by the matching unit 21. Have been. The automatic dictionary generation unit 24 includes a parameter setting unit 18 and a dictionary image processing unit 19.

FIG. 2 shows an example of multi-font registration in the dictionary memory 20 for the character "9", and the font f0 is a basic font in which the characters on the captured image are registered as they are. A font f1 rotated right by image processing, a font f2 rotated left, an enlarged font f3, a reduced font f4, a font f5 rotated right after enlargement, a font f6 rotated left after enlargement by image processing,
The dictionary automatic generation unit 24 automatically generates a font f7 rotated right after the reduction and a font f8 rotated left after the reduction,
These registered fonts can be variously set by the parameter setting unit 18.

FIG. 3 shows an example of contents set by the parameter setting unit 18 and a setting range thereof, and FIG. 4 is an explanatory diagram of the contents set. The setting contents in FIG. 3 include an X-direction enlargement ratio xmag1, a Y-direction enlargement ratio ymag1, an X-direction reduction ratio xmag2, Y
It consists of a direction reduction ratio ymag2, a rotation angle rotate, the number n of fonts to be automatically generated, and a content specification for specifying what kind of image processing (enlargement, reduction, rotation) has been performed for each of the n fonts. . This content specification is performed by using numbers. For the basic font (font created by capturing with a camera or the like), <0> is rotated clockwise and <1> is specified.
Is left rotation, <2> is expansion, <3> is reduction, <4> is right rotation after expansion, <5> is left rotation after expansion, <6> is right rotation after reduction, and <7> is left after reduction. The rotation is specified.

The details of the enlargement, reduction, and rotation are shown in FIG. Now, the frame 25 in FIGS. 4A and 4B is a basic font of the character “9”, the height is H, and the width is W.
Expressed by The frame 26 in FIG. 4A is the font after the enlargement process, and its height is represented by H1 and its width is represented by W1. H1 = H
× ymag1, W1 = W × xmag1. The frame 27 in FIG. 4B is the font after the reduction process, and its height is H.
2. If the width is represented by W2, H2 = H × ymag2, W2 = W ×
xmag2. H1 in FIG. 4A represents the upper and lower enlargements, and h1 = (H1−H) / 2. In addition, w1 represents the left and right enlargement, and w1 = (W1−W) / 2. Similarly, FIG.
H2 and w2 are reductions, and h2 = (H−H2) /
2, w2 = (W−W2) / 2. FIG. 4C shows the clockwise rotation, in which the alternate long and short dash line 28 is the central axis of the basic font before the processing, and the alternate long and short dash line 29 is the central axis after the rotation processing. r1 is a rotation angle, and r1 = rotate. This rotation can be performed for any of the basic font and the enlarged / reduced font.

In order to generate the eight fonts f1 to f8 shown in FIG. 2 by the parameter setting unit 18 shown in FIG. 3, the automatic generation number n is 8, and the content designations <0> to <7 for the fonts f1 to f8. > In this order. However, it is assumed that desired values for the enlargement / reduction ratio, the rotation angle, and the like are given in the corresponding columns of the parameter setting unit.

Next, the operation of the apparatus shown in FIG. 1 will be described.
1 receives a detection signal from the sensor 13 indicating that the inspection target 11 has come to a predetermined position, starts the video camera 12, and the video camera sends the inspection target to the image processing device 23. Grayscale image data (for example,
(255 gradations). The video capturing unit 15 receives this image data and stores it in a predetermined image memory 16.

When the dictionary registration is performed, the dictionary registration unit 17 is activated, and performs the registration processing according to the flow shown in FIG. First, in step ST1, a selected area of a grayscale image input from a video camera or the like is registered in the dictionary memory 20 as a basic font as it is. In the next step ST2, the parameter i indicating the font number is set to "0".
Then, in step ST3, it is compared with the automatically generated number n set by the parameter setting unit 18 in FIG. If they are equal, the process ends. If they are different, in step ST4, the image processing is performed according to the contents set for each font by the parameter setting unit 18 in FIG.
Do with. For example, if the content specification <5> is specified as the font i + 1, the basic font is enlarged at the specified enlargement ratio, and then a font rotated left by the specified angle is generated. This generated font is stored in step ST5.
To register in the dictionary memory 20 as a pair with the parameter i + 1. Then, in step ST6, the value of the parameter i is set to 1
Increase and repeat the next font generation.

In the case of inspecting the character shape, step ST
The pattern matching unit 21 performs pattern matching between the pattern image of the check character captured in the image memory 16 at 16 and the font stored in the dictionary memory 20. When a plurality of dictionaries are registered for one character as a multi-font, this pattern matching is performed for all the fonts. This is the same as in the prior art, and the result determination unit 22 determines whether the shape of the character is correct with reference to the degree of coincidence, which is the result of the pattern matching operation performed by the matching unit 21.

[0018]

As described above, according to the present invention,
Fonts that have undergone image processing, including enlargement and reduction, are automatically generated based on the values set in parameters, so that the number of steps required for dictionary registration can be reduced, and shape inspection utilizing multi-fonts can be performed easily. There is an effect that can be.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration example of a character inspection device according to the present invention.

FIG. 2 is a diagram illustrating an example of a multi-font.

FIG. 3 is an explanatory diagram of setting contents of a parameter setting unit.

FIG. 4 is an explanatory diagram of parameters in FIG. 3;

FIG. 5 is a flowchart showing a dictionary registration process which is a feature of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 11 Inspection object 12 Video camera 13 Sensor 14 Video capture signal generation part 15 Video capture part 16 Image memory 17 Dictionary registration part 18 Parameter setting part 19 Image processing part 20 Dictionary memory 21 Collation part 22 Result judgment part 23 Image processing part 24 Dictionary Automatic generation unit

Claims (3)

[Claims] 1. A dictionary memory in which a plurality of fonts are registered for one character, and a character inspection is performed by comparing and comparing a pattern of an input inspection target character with each of the fonts registered in the dictionary memory. A character setting device configured to set parameters for transforming a font of a standard character captured and created by the character image input means by image processing including at least enlargement and reduction processing; Font image processing means for performing image processing including at least enlargement and reduction in accordance with the parameters set by the means, generating a transformed font, and registering the transformed font in the dictionary memory. Character inspection device. 2. The method according to claim 1, wherein the parameter setting unit includes at least an enlargement ratio of the standard character in a vertical direction, an enlargement ratio in a horizontal direction,
2. The character inspection apparatus according to claim 1, wherein the character inspection device is a means for setting a parameter indicating a vertical reduction ratio and a horizontal reduction ratio. 3. A dictionary automatic generation method for a character inspection apparatus for registering a plurality of fonts of the same character in a dictionary for character inspection, wherein a reference font of the character is generated from image data taken in by a character image input means. Further, another character font is generated for the character by performing image processing including at least one of enlargement and reduction processing on the reference font, and registered in a dictionary. Generation method.