JPH07159340A - Inspection apparatus for printed object - Google Patents

Abstract

PURPOSE:To provide an inspection apparatus for a printed object, which can inspect the quality of a printed character with high accuracy even when an object to be measured, on which a character has been printed, is being conveyed. CONSTITUTION:Light is projected to an object S to be measured from a light source 21, its reflected light is detected by an image input sensor 22, its binarized data is stored in an image memory 16, and light-and-shade image data is stored in an image memory 17. A CPU 11 detects the position and the inclination of a ruled-line frame on the basis of an image stored in the image memory 16, the dislocation amount of the ruled-line frame is compared with a criterion value stored in a dictionary pattern memory 18, an extracted image as a whole is position-corrected including its rotation, the inclination of a character string is detected from the character inspection region of the corrected image, the rotation of the character string is corrected, every character is cut off with reference to the corrected image of the character string by means of a density projection method, and a printed character is extracted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for inspecting printed matter, and for inspecting a quality such as a manufacturing date stamped during transportation in a specific area surrounded by a ruled line in a packaging carton or the like during transportation. Inspection device for various printed matter.

[0002]

2. Description of the Related Art Conventionally, as a method and apparatus for inspecting the quality of printed characters (character missing, etc.), for example, Japanese Patent Laid-Open No. 2-14379 or Japanese Utility Model Laid-Open No. 1-160558.
There is one disclosed in the publication. All of these are simply judged as good or bad simply by comparing with standard character pattern information.

However, there are various types of printing defects depending on the cause of the printing defects. Although the person recognizes the type and judges the quality from the judgment standard according to the standard, it is obvious that the processing content is far from the visual judgment in the conventional uniform comparison method. Therefore, Japanese Laid-Open Patent Application No. 4-286082 proposes a method of restricting defect items in advance and processing the characters so that the extracted characters can be compared for each selection item. In this method, when extracting a character, the position image of the character is cut out into a rectangle from the density histogram by projection in the vertical and horizontal directions, the character is determined to be a character in the frame, and the image in the frame is identified as each defect. It is processed for comparison.

[0004]

However, for example, when inspecting the quality of characters, such as the date of manufacture, which is imprinted in a designated area within a ruled line during conveyance, a positional deviation at the inspection position of the printing object is caused. If it always occurs, the extracted character will be judged to be misaligned even if it is imprinted at the regular position in the ruled line frame. Further, when a character is detected to be tilted, the character area cannot be cut out by the above-described density projection method. further,
In the above-mentioned proposal, for example, it is not possible to extract the density of characters, and it is not possible to select the degree of thinning or thickening of characters, and there is a problem in the reliability of character quality inspection.

Therefore, the main object of the present invention is to enable highly reliable character extraction, greatly enhance the selection items of character quality as compared with the conventional one, and to perform a highly reliable inspection of a printed matter. Is to provide.

[0006]

SUMMARY OF THE INVENTION The present invention is a printed matter inspection apparatus, which collects characters by reading characters in a specific area including a ruled line frame, and extracts coordinates of the ruled line frame from the collected image. First correction means for extracting the deviation amount of the ruled line from the reference value and correcting the rotation of the entire collected image, and the coordinates of the character string from the corrected image, and extracting only the character area image from the inclination amount of the character string. Second rotation correction means for rotation correction, extraction means for extracting a circumscribed rectangle of a character from the corrected image, dictionary storage means for storing reference character information in advance, and a portion of the extracted circumscribed rectangle character. To determine whether or not a specific omission has occurred by comparing with a reference value stored in the dictionary storage means
Processing means, second processing means for judging the size of the extracted circumscribed rectangular characters by comparing with the reference value stored in the dictionary storage means, and a dictionary of the density of the extracted circumscribed rectangular characters. Third processing means for making a judgment by comparing with a reference value stored in the storage means, and relative position deviation of the extracted ruled line in the circumscribed rectangle from the reference position in the dictionary storage means. Fourth processing means for making a judgment by comparing with a stored reference value, and fifth processing means for making a judgment by comparing the dirt of the extracted circumscribed rectangle character with a reference value stored in the dictionary storing means. And a stain for the background image between the corrected ruled line image and the corrected character area is extracted and determined.
And the sorting means for sorting the collected character states by comprehensively judging the results from the first to sixth processing means.

[0007]

In the printed material inspection apparatus according to the present invention, when inspecting a character including a ruled line frame, first, the position and inclination of the ruled line frame are detected, and the positional deviation amount of the ruled line frame is compared with a reference value and extracted. Position correction including rotation is performed on the entire image, then the inclination of the character string is detected from the character inspection area of the corrected image, the character string is rotated and corrected, and each character is corrected on the corrected character string image. It is possible to extract characters with high reliability by performing the density projection method for clipping, and greatly enhance the selection items of character quality compared to the conventional method, and to perform highly reliable inspection.

[0008]

1 is a block diagram showing the overall structure of the present invention. In FIG. 1, a CPU 11 performs overall control, and a program memory 14 and a working memory 1 are provided via a data bus 12 and an address bus 13.
5, the image memories 16 and 17, the dictionary pattern memory 18, and the output circuit 19 are connected.

On the other hand, the object to be measured S is irradiated with light from the light source 21, and the reflected light from the object to be measured 3 due to the irradiation light is coupled to the image input sensor 22. Image input sensor 22
Is output to the image input circuit 23 as character image data. The output of the image input circuit 23 is a binarization circuit 24.
And the image memory 17 are provided. The binarization circuit 24 outputs the output of the image input circuit 23 according to an internal set threshold value.
The value is converted and given to the image memory 16. At the same time, the grayscale image output from the image input circuit 23 is directly stored in the image memory 17 without passing through the binarization circuit 24. The program memory 14 stores a processing program for the CPU 11 to perform overall control, various control data, reference data, and the like.

Here, as the processing program, for example, based on the storage contents of the image memories 16 and 17, the position correction with respect to the reference value of the image, the character cut-out of the corrected image, and the cut-out character can be compared with each dictionary pattern. , Feature quantity extraction for each selection item, comparison with dictionary patterns regarding missing or bleeding of characters, or masking the rectangular area of the cut-out character, and a label image showing whether the background part between the ruled line and the mask area is dirty The conversion is performed to make a determination based on the connection state, and further the above comparison results are comprehensively determined to determine whether or not there is a printing defect in the character image, and to select a printing defect.

The dictionary pattern is a standard pattern to be compared in the above-mentioned processing, and when a reference print character is formed on the image input sensor 22 under the same condition as the object to be inspected, it is input from the image input circuit 23. It This standard pattern is processed into the following standard dictionary pattern and serves as a reference for each comparison item.

The standard patterns described above include a position dictionary pattern, a middle point dictionary pattern, a density dictionary pattern, and an outer point dictionary pattern. The position dictionary pattern is to determine the positional deviation of the object to be measured from the ruled line frame of the input image, and after correcting the coordinate of the image, to make it possible to stably cut out characters even when the character string is slightly inclined. The amount of inclination of the character string is extracted, the character string image is rotated and corrected, and then the extracted character is compared. Therefore, by comparing with this dictionary pattern, it is possible to finally detect the displacement and inclination of the character from the normal printing position. In addition,
The rotation correction is performed within a range in which an allowable value is set in advance.

The midpoint dictionary pattern is to be compared with the extracted character after being corrected as described above. As the middle point dictionary pattern, X, Y
Scanning is performed in each direction, and the midpoint of consecutive character coordinates on each scanning axis is extracted. Therefore, by comparing with the dictionary pattern, it is possible to extract the above-mentioned defects such as chipping, blurring, and voids of the character image.

FIG. 2 is a diagram showing characters for a standard dictionary as an example of the standard dictionary pattern, and FIG. 3 is a diagram showing an example of a midpoint dictionary pattern.

FIG. 4 is a diagram showing an example of the density dictionary pattern. This density dictionary pattern is to be compared with the extracted character after being corrected as described above. As this density dictionary pattern, an image input for a standard dictionary is output to a display device, and while watching the image, by teaching, a plurality of lines are overwritten on each character image at arbitrary intervals. The coordinates of the overlapping portion of are extracted. The density of the character can be extracted by comparison with this standard pattern. Furthermore, by counting the number of images on the line, it is possible to extract the amount of thinning or thickening of a character that cannot be extracted by the above-mentioned midpoint dictionary pattern. However, the density extraction data is input from the corrected image memory 16, and the character thinning extraction data is input from the corrected image memory 17.

FIG. 5 is a diagram showing an example of the external point dictionary pattern. After correcting the external point dictionary pattern as described above,
This is to be compared with the extracted character, and the inside of the cut out character frame is converted into a label image and expansion processing is performed to extract the character coordinates where the expansion image and the pre-expansion image do not overlap. Therefore, by comparison with this standard pattern, it is possible to extract defects such as bleeding, flow, and thickening of the character image.

As described above, when the count result is judged to be the allowable value or more by comparison with the dictionary pattern as described above, it is detected as missing or bleeding.

On the other hand, in order to detect the contamination of the background portion (inside the ruled line frame), the character coordinates are masked and the background portion in the ruled line frame is converted into a label image. Contamination is extracted from the number of connected images extracted in this process. FIG. 6 is a diagram showing an example of the label image converted to connect the background partial images.

The types of printing defects include, for example, "chip", "void", "cut", "bleed", "thin",
"Blurring", "thickness", "double printing", "misalignment",
1 of "tilt", "background stain", and "dirt stain"
One type is selected.

FIG. 7 is a flow chart for explaining the operation of one embodiment of the present invention. Next, FIGS.
A specific operation of the embodiment of the present invention will be described with reference to FIG. Light is emitted from the light source 21 to the object to be measured S, the reflected light is detected by the image input sensor 22, and the output thereof is taken into the image input circuit 23 as character image data. The output of the image input circuit 23 is stored in the image memory 17 as grayscale image data, binarized by the binarization circuit 24, and stored in the image memory 16. CP
U11 extracts the coordinates of the ruled line frame from the image stored in the image memory 16 in step (abbreviated as SP in the figure) SP1, and corrects the position of the image data in step SP2. At this time, the coordinate data of the unnecessary portion outside the ruled line frame is set to zero. Next, the CPU 11 obtains the coordinates of the character string from the image data of the character inspection area. Step SP
In 3, the inclination amount of the character string is obtained from the coordinate data. The inclination amount is generally obtained by determining the principal axis angle from the center of gravity of the character string, but the method is not limited to this method.

Next, in step SP4, the CPU 11 performs rotation correction of the calculated tilt amount on the character string image. This correction is performed only when a tilt amount within a preset allowable tilt angle is extracted. Next, in SP5, a rectangular frame is cut out for each character from the corrected character string. This work is preferably performed by projecting the image density of each of the X and Y axes, but is not limited to this method. In this way, the extracted rectangle size is compared with the number of characters in the standard dictionary in step SP6. If the result is significantly different from the reference value, it is determined that the printing process is abnormal or the inspection condition is abnormal, and the process proceeds to step SP12. move on. Next, when the number of extracted rectangular areas is different from the number in the standard dictionary, it is selected as dirt between characters or dirt on the background area of the inspection area.

Next, the characters in the rectangular area are determined in step SP.
In 8 to SP10, the above-mentioned respective features are extracted and sequentially compared with the standard dictionary pattern to be selected. That is,
In step SP8, it is compared with the middle point dictionary pattern, in step SP9 it is compared with the density dictionary pattern, and in step SP10 it is compared with the outer point dictionary pattern. The reference value of each selection item is preferably set for each character. Next, in step SP11,
The number of connected images in the background portion inside the ruled line is counted, and the background stain is extracted. Finally, in step SP12, comprehensive determination of the extracted information is performed, and in step SP13, the result is output to the output circuit 19. As a result, when the series of processes is completed, the process returns to step SP1 and the above process is repeated. However, in the above-described process, in order to determine an abnormality or an inspection condition abnormality, the continuity of the result determined to be defective in step SP14 is confirmed before returning to step SP1, and if continuous, the process is abnormal in step SP15. If there is, or it is determined that the inspection condition setting is abnormal, and the message is output to the output circuit 19, the process is interrupted.

In the above-described embodiment, an example in which a character to be printed at a predetermined position including a ruled line frame is inspected during conveyance is shown. However, the present invention is not limited to this, and it is possible to inspect in a stationary state, for example. Of course, the characters can also be easily applied to the inspection of kanji, katakana, hiragana, and alphabets.

[0024]

As described above, according to the present invention, when a character including a ruled line frame is inspected, the position and inclination of the ruled line frame are detected, and the positional deviation of the ruled line frame is compared with a reference value. Position correction including rotation is performed on the entire extracted image, then rotation of the character string is corrected by detecting the inclination of the character string from the character inspection area of the corrected image, and each corrected character string image is corrected. Since the character is cut out by the density projection method, the quality of the printed character can be inspected with high accuracy even when the measured object on which the character is printed is being conveyed.

[Brief description of drawings]

FIG. 1 is a block diagram showing the overall configuration of an embodiment of the present invention.

FIG. 2 is a diagram showing an example of characters for a standard dictionary pattern.

FIG. 3 is a diagram showing an example of a midpoint dictionary pattern.

FIG. 4 is a diagram showing an example of a density dictionary pattern.

FIG. 5 is a diagram showing an example of a rotating dictionary pattern.

FIG. 6 is a diagram showing an example of a label image of a background portion.

FIG. 7 is a flow chart for explaining the operation of the embodiment of the present invention.

[Explanation of symbols]

 11 CPU 12 data bus 13 address bus 14 program memory 15 working memory 16 position image memory 17 grayscale image memory 18 dictionary pattern memory 19 output circuit 21 light source 22 image input sensor 23 image input circuit 24 binarization circuit

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G06K 9/32 9/34 9/56

Claims (1)

[Claims] 1. A collecting unit that reads and collects characters in a specific area including a ruled line frame, coordinates of the ruled line frame are extracted from an image collected by the collecting unit, and a deviation amount from a reference value of the ruled line is extracted. A first correction means for rotationally correcting the entire collected image; a coordinate of a character string extracted from the image corrected by the first correction means; and a rotational correction for only the character area image based on the inclination amount of the character string. 2 rotation correction means, extraction means for extracting a circumscribed rectangle of a character from the image corrected by the second rotation correction means, dictionary storage means for storing reference character information in advance, and extraction by the extraction means First processing means for determining a partial omission of the extracted circumscribed rectangular characters by comparing with a reference value stored in the dictionary storage means; and size of the circumscribed rectangular characters extracted by the extraction means. The dictionary memory Second processing means for making a judgment by comparing with a reference value stored in the means, and comparing the density of the characters in the circumscribed rectangle extracted by the extracting means with the reference value stored in the dictionary storage means. Third processing means for judging, and relative position deviation from the reference position of the ruled line corrected by the first correction means for the positional deviation of the circumscribing rectangle extracted by the extraction means is stored in the dictionary storage means. A fourth processing means for making a determination by comparing with a reference value stored in the dictionary; and a fourth processing means for making a determination by comparing the dirt of the characters in the circumscribing rectangle extracted by the extracting means with a reference value stored in the dictionary storage means. 5) processing means, 6th processing means for extracting and determining dirt on the background image between the ruled line image corrected by the first correction means and the character area corrected by the second correction means The first to sixth processes Comprehensive evaluation to the inspection apparatus of a printed material having a selection means for selecting the state of characters collected by the collecting means results from the means.