JPH10269357A - Method for inspecting printed matter - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically inspect a printed matter without performing positioning (registering) manually even if the position of a printed matter to be inspected is roughly inputted. SOLUTION: A sample image that becomes a comparison reference is stored in reference image memory (S1), a positioning design is selected from the sample image and binatrized, an center of gravity position coordinate is calculated and the positioning design is stored in model image memory (S2 to S5). Next, a print image to be inspected is stored in check image memory, a positioning design that agrees to a model image is retrieved by using a correlation coefficient method, and a center of gravity position coordinate is calculated (S6 to S8). S2 to S9 are repeated, the center of gravity position coordinate of other positioning design is calculated, affine transformation is performed to the check image memory based on the center of gravity position coordinates of the two points, the sample image is compared with the print image to be inspected in a pixel unit and the different points between them are retrieved (S9 to S11).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an inspection method for detecting dirt, chipping, and incorrect characters in printed matter. Especially,
Even if the position of the test print is roughly input, the difference between the two can be searched by correcting the address of the test image memory of the test print image and comparing the image information for each address of the reference image memory of the sample image. The present invention relates to a method for inspecting printed matter.

[0002]

2. Description of the Related Art When printing a printed matter, a proofreading operation is performed. This is a task of checking the sample and the printed matter at the beginning of printing while comparing them over the entire surface. The emphasis here is on checking that there is no problem in the character part.
For example, in the field of package printing, a description of a product or the like is printed on a packaging material. In particular, if a medicine becomes dirty, chipped, or has a wrong character, a consumer may use the product incorrectly, which is extremely dangerous.

[0003] However, the inspection work imposes a heavy load on the operator due to the small size of the characters, the large number of multi-faced prints, and the like, and oversight rarely occurs.

[0004] In recent years, an apparatus for automatically inspecting a printing machine or a rewind inspection machine equipped with a camera and an image processing apparatus has appeared, but such an apparatus also performs 100% inspection at a high printing speed. Therefore, the resolution of the camera itself, which is an input device, is low, which is not suitable for such an inspection.

[0005] In view of this, an inspection apparatus has been developed which prepares a high-resolution image input apparatus such as a flatbed scanner, inputs a sample and a printed material to be inspected as an image, and performs image processing to detect a difference between the two. .

However, in such an apparatus, it is impossible to input an image by locating the sample and the inspection product in exactly the same manner. Then, the input sample image is displayed on a TV monitor or the like, and an appropriate pattern is selected in advance by the operator for alignment from the image, and one register is registered so that the sample image matches the inspection print image. You have to adjust each time.

Here, the "pattern suitable for positioning" is a line drawing pattern in which the contrast between the ground color of the paper and the printing color is clear and there is no shape similar to the periphery. (Hereinafter, referred to as the alignment pattern)

When at least two corresponding position coordinates are obtained from the images of the sample and the inspection product, the expansion and contraction and inclination of the paper between the sample and the inspection product can be calculated. By affine-transforming either the sample or the sample image using these values, it is possible to compare both images for each pixel.

The method of inputting the image of the inspection printed matter, and selecting the alignment pattern and aligning the position each time the input is performed, cannot cope with a large number of inspected sheets.

In particular, in the field of package printing in which multiple imposition printing is performed, the same printed matter must be inspected at once in tens of thousands to hundreds of thousands. Therefore, in order to automate the inspection and make the inspection unattended, it is necessary to perform a batch process in which an automatic document feeder or the like is attached to a flatbed scanner and a large number of inspection products are continuously image-inputted and inspected. To achieve this, the alignment of the inspection product has been a bottleneck.

[0011]

SUMMARY OF THE INVENTION The present invention has been proposed in view of such a situation. In order to enable inspection even if positioning is rough, a pattern for alignment is prepared by an operator. Is selected in advance, and from the image of the inspection printed material, a corresponding alignment image is selected, and
Provide a method of inspecting a printed material that performs image processing by automatically superimposing both image positions by performing coordinate transformation.

[0012] Thus, each time an inspection product image is input, a position corresponding to the alignment pattern selected by the sample is designated on the image of the inspection printed material, and the operation of aligning is unnecessary, and the load on the operator is reduced. Aim. Further, it is possible to perform batch processing in which an automatic document feeder or the like is attached to the flatbed scanner and a large number of inspected products are continuously image-inputted and inspected, thereby improving work efficiency.

[0013]

According to the present invention, a picture of a printed matter is read by an image input device, and the read picture information of each pixel is compared with reference picture information of a corresponding pixel inputted in advance. (1) A step of reading a sample image from an image input device and storing the sample image in a reference image memory of a memory unit of the image processing device. (2) a step of displaying the sample image on an image display device and selecting a pattern inspection area including a pattern for alignment; (3) A step of cutting out a rectangular area including only the alignment pattern from the pattern inspection area, creating a gradation histogram of the rectangular area, and binarizing the alignment pattern. (4) a step of performing contour tracing on the binarized alignment pattern to obtain a circumscribed rectangle and calculating a barycentric position. (5) a step of storing, as a model image, the positioning picture surrounded by the circumscribed rectangle in a model image memory of a memory unit of the image processing apparatus. (6) A step of reading the inspection print image from the image input device and storing it in the inspection image memory of the memory unit of the image processing device. (7) a step of searching a pattern inspection area of the inspection printed matter for an alignment pattern that matches the model image by using a correlation coefficient method. (8) A step of creating a gradation histogram of a pattern for alignment of an inspection print and then binarizing the gradation histogram. (9) A step of calculating the coordinates of the center of gravity of the alignment pattern of the inspection print. (10) A step of repeating the steps (2) to (9) to calculate the barycentric position coordinates of another alignment pattern. (11) A step of performing affine transformation on the inspection image memory based on the coordinates of the center of gravity of the two points. (12) A step of comparing, on a pixel-by-pixel basis, the test print image image and the sample image, which have been coordinate-transformed by the affine transformation, to search for differences therebetween. This is a method for inspecting printed matter comprising:

[0014]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a processing flow of the inspection processing method of the present invention, and FIG. 2 shows a hardware configuration.
An image processing device (1) and an image input device (2) including a CPU unit, an image memory unit, an image input unit, and an image display unit, a TV monitor for image display (3), a keyboard (4), a mouse (5) ) Is used. In this embodiment, the main processing is performed by software using a flatbed scanner as the image input device (2).

FIG. 3A shows an example of the whole sample image, FIG. 3B shows a part of the inspection print image including a pattern for alignment, and FIG. 3C shows the inspection image of FIG. This shows a state where the printed image is placed obliquely.

The process of detecting the difference between the sample and the inspection product is basically performed by a method of comparing corresponding pixels of each image. However, as shown in FIG. 3 (B), it is difficult to accurately place a sample or an inspection product horizontally on a flatbed scanner. Gets worse. For this reason, simply comparing pixels at the same coordinates in an input image results in comparisons between different positions in the image, and is detected as a difference over the entire surface.

Therefore, image processing is performed in which one of the images is rotated and tilted at the same angle as the other image. At the same time, it is necessary to perform the size conversion process in consideration of the expansion and contraction of the paper, and for that purpose, it is necessary to determine the degree of the inclination and the expansion and contraction of the size between the sample and the inspection product. For this purpose, the following processing is performed.

Sample image (10) from image input device (2)
Is input and stored in the reference image memory of the memory unit of the image processing apparatus (1). (Processing step S1) This image is displayed on the image display TV via the image display unit.
Display on the monitor (3). Here, the operator selects an appropriate pattern for the positioning. (Processing step S
2) The "positioning pattern" is a line drawing pattern in which the contrast between the ground color and the printing color of the paper is clear and there is no shape similar to the periphery. In the example of this printed matter, description will be made assuming that “La” (12) is selected as the alignment pattern.

It is confirmed whether or not this "la" is suitable as a positioning pattern. That is, FIG. 4 shows a processing flow for confirming whether or not there is an approximate pattern near “la”.

The operator cuts out a rectangular area of the image including the alignment pattern. In the cutout process, a tone value histogram of an area of a predetermined size centered on coordinates specified by the mouse by the operator is created, and the image is binarized using a discriminant analysis method or the like. (Processing step S3) Then, the operator designates a pattern for positioning with a mouse, performs contour tracing, obtains a rectangle circumscribing the pattern for positioning, and simultaneously calculates the position of the center of gravity. (Processing step S
4)

The registration pattern surrounded by the circumscribed rectangle is stored as a model image (13) for pattern search. Then, it is stored as a pattern search area (11) surrounded by a rectangle centered on the model image. (Processing step S5)

The size of the pattern search area (11) is determined in advance according to how rough the operator is allowed to set sheets. In other words, the size is increased when it is assumed to be placed roughly, and the size of the pattern search area is reduced when it is placed at the same position as the sample. Here, if a pattern having a shape similar to the pattern selected by the operator exists in the pattern search area (11), an alarm message is displayed on the screen. In order to prevent the possibility that the approximate pattern cannot be distinguished from the alignment pattern, it is necessary to select another alignment pattern.

The determination as to whether or not an approximate shape pattern exists can be made in the same manner as in a pattern search process using a correlation coefficient described later.

Next, the image of the inspection product is input and stored in the inspection image memory of the memory unit of the image processing apparatus (1). (Processing Step S6) Here, in the conventional method, it is necessary to display an image in the same manner as the sample and to designate and align a position corresponding to a picture position designated by the sample by the operator. By doing so, a positioning process is performed.

The position of the previously registered model image (13) is searched from the image at the position corresponding to the previously stored pattern search area from the inspection image memory. In the search, the correlation coefficient between the model image (13) and the image area of the same size of the test print image is calculated starting from the upper left corner of the pattern search area (21) of the test print. (Processing step S
7) In this embodiment, as shown in FIG.
And a correlation coefficient calculation area (23) having the same size. When the calculation of this position is completed, the correlation coefficient calculation area (23) is shifted rightward by one pixel to obtain a correlation coefficient at that position.

As described above, the position of the correlation coefficient calculation area (23) is sequentially moved, and the correlation coefficient with the model image (13) is calculated over the entire pattern search area (21) of the inspection print image. After obtaining the correlation coefficient of the entire pattern search area (21) of the inspection product image, the position where the value becomes the largest is found. At that time, if the maximum value of the correlation coefficient is smaller than a predetermined value, it is determined that the search for the model image has failed,
The inspection product is set on the flatbed scanner again, and a message for instructing image input is displayed.

Assuming that the model image data is Mi and the image data of the correlation coefficient calculation area is Ii, the correlation coefficient r at the position (u, v) in the pattern search area can be obtained by the following equation.

[0028]

(Equation 1)

Here, N is the number of pixels included in the model. It can be said that both patterns are more similar as the correlation coefficient is larger.

If the maximum value of the correlation coefficient is larger than a predetermined value, an image pattern matching the model image (13) exists at that position, and the barycentric coordinates are obtained. (Processing Step S8) By this processing, it becomes possible to automatically select a pattern position corresponding to the pattern position specified by the operator in the sample image from the inspection image.

When such processing is performed at least two points, two coordinate values of the same picture position of the sample image and the inspection print image can be calculated. (Processing step S9) In the present embodiment, "2" is used as the overlapping pattern as another point.
Was selected.

In the affine transformation, the coordinates (x, y) can be transformed into coordinates (x ', y') by a transformation formula of x '= ax + by y' = cx + dy. Where a,
The parameters of b, c, d can be obtained if two points (x, y) and (x ', y') are given. A, b, c, d can be calculated using the coordinate values of the two points at the same picture position of the sample and the inspection product. By performing affine transformation using this value, the subsequent comparison processing can be performed in a state where the inclination of the sheet and the expansion and contraction of the sheet when the operator sets the sheet on the flatbed scanner are corrected. (Processing Step S10) The process of comparing the sample and the inspected product in pixel units after correcting the inclination and expansion / contraction to find the difference between the two is performed by a generally used image processing method. (Processing step S
11)

[0033]

According to the present invention, even when the operator performs rough positioning of the test print on the flatbed scanner, if the operator matches the sample positioning image, the test print positioning image is obtained. , And an image processing for comparing the two based on the position coordinates makes it possible to perform an inspection method for detecting a stain, a chip, a character difference, or the like of a printed matter.

That is, this is an inspection method that does not require a person to align a picture in order to compare the two.
The inspection process can be automated and unmanned. Batch processing, such as attaching an automatic document feeder or the like to a flatbed scanner and continuously inputting and inspecting a large number of inspected products, is possible. The effect is particularly large in some cases.

[Brief description of the drawings]

FIG. 1 is a processing flowchart of a printed matter inspection method of the present invention.

FIG. 2 is a diagram illustrating a hardware configuration of an embodiment of the present invention.

FIG. 3A is a sample image of an inspection standard. (B)
FIG. 4 is a partial view including a pattern for alignment of an inspection print. (C) is a figure which shows the state in which the inspection printed matter of (B) was inclined.

FIG. 4 is a diagram showing a processing flow for cutting out a rectangular area used as a model image from an image selected by an operator.

FIG. 5 is a diagram illustrating a manner in which a correlation coefficient with a model image is calculated while sequentially shifting the position of the model image from a pattern search area of an inspection print image.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Image processing apparatus 2 ... Image input apparatus 3 ... Image display TV monitor 4 ... Keyboard 5 ... Mouse 10 ... Sample image used as an inspection reference 11 ... Sample image pattern search area 12 ... Sample image positioning pattern 13 ... A rectangular image (model image) including a pattern for positioning 14... A pattern search area of another sample image 15... A pattern for positioning of a sample image 20... An image of an inspection print 21. Image alignment pattern 23: Correlation coefficient calculation area

Claims (1)

[Claims] An image input device reads a picture of a printed matter,
In the inspection method for comparing the read picture information of each pixel with the reference picture information of the corresponding pixel which is input in advance to detect a stain, a chip, a character difference, or the like of a printed matter, Reading a sample image and storing the sample image in a reference image memory of a memory unit of the image processing apparatus. (2) a step of displaying the sample image on an image display device and selecting a pattern inspection area including a pattern for alignment; (3) A step of cutting out a rectangular area including only the alignment pattern from the pattern inspection area, creating a gradation histogram of the rectangular area, and binarizing the alignment pattern. (4) a step of performing contour tracing on the binarized alignment pattern to obtain a circumscribed rectangle and calculating a barycentric position. (5) a step of storing, as a model image, the positioning picture surrounded by the circumscribed rectangle in a model image memory of a memory unit of the image processing apparatus. (6) A step of reading the inspection print image from the image input device and storing it in the inspection image memory of the memory unit of the image processing device. (7) a step of searching a pattern inspection area of the inspection printed matter for an alignment pattern that matches the model image by using a correlation coefficient method. (8) A step of creating a gradation histogram of a pattern for alignment of an inspection print and then binarizing the gradation histogram. (9) A step of calculating the coordinates of the center of gravity of the alignment pattern of the inspection print. (10) A step of repeating the steps (2) to (9) to calculate the barycentric position coordinates of another alignment pattern. (11) A step of performing affine transformation on the inspection image memory based on the coordinates of the center of gravity of the two points. (12) A step of comparing, on a pixel-by-pixel basis, the test print image image and the sample image, which have been coordinate-transformed by the affine transformation, to search for differences therebetween. A method for inspecting printed matter, comprising: