JPH09265192A - Registration method and device therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To exactly execute registration by subjecting the positions of the detected register marks of an image plate to affine transformation with respect to the register marks of a reference image plane. SOLUTION: An image forming section 19 to which the image signals supplied in a frame memory 14 determines the frequencies at which respective densities are detected with each of respective pixels. The section binarizes the frequencies by each of the pixels, imparts address signals thereto and stores the signals as characteristic images into a random access memory(RAM)21. The images of which the shapes have the known register marks are processed by an image input device 13. The image signals of a frame memory 14 are immediately binarized and the register characteristic images 24 are obtd. and are stored in a RAM 25 for register. A register mark detecting section 26 scans the images while executing differential computation of the characteristic image signals and the register characteristic signals and outputs the positions where the results of the computation is minimized to a register mark position display section 27. These positions are subjected to the affine transformation with respect to the position of the register marks of the image plate which is the reference and are registered to the respective image plates.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention finds the position of a register mark of an image plane with respect to a register mark of an image having a register mark of a known shape, and determines the reference image plane and other image planes. Registration method and apparatus.

[0002]

2. Description of the Related Art In ordinary multicolor printing, various colors such as black, indigo, crimson, and yellow are printed on a sheet of paper to reproduce various colors. At this time, it is necessary that the relative printing positions of the respective colors are matched on the paper surface as much as possible, and the respective colors are printed without being found. It is said that the tolerance of the coincidence is ± 50 μm or less.

Conventionally, the registration between each color in multicolor printing is
This is done manually, and a human being visually confirms the misregistration of each color by trial printing, and the color is registered by a register adjusting device. However, the registration work has a tolerance of ± 50.
Since a strict accuracy of μm or less is required, it has been done by a skilled worker, and a heavy load is applied to the worker.

For this reason, recently, in multicolor printing,
Devices have been developed for automatic registration. For example, Japanese Unexamined Patent Publication No. 4-76555 discloses a method of performing automatic registration using a register mark having at least two line segments that intersect at right angles. In this method, the center position is specified by tracing a register mark having two line segments intersecting at right angles,
The registration was done and was intended only for specific register marks having at least two line segments that intersect at right angles. Therefore, there is a problem in that it is not possible to detect the position of various kinds of register marks currently in use and register them. Also,
Since the image of the register mark is different for each color image, there is a problem that its shape is difficult to read.

[0005]

SUMMARY OF THE INVENTION The object of the present invention is to solve such problems of the prior art, and regardless of the shape of the register mark used,
It is an object of the present invention to provide a method and apparatus for accurately detecting the position of a register mark and registering it regardless of the state of an image.

[0006]

According to the present invention, an image having a register mark of a known shape is divided in a horizontal direction and a vertical direction to provide a plurality of pixels, and each pixel has two register marks. The image version for which the register mark is to be detected is converted into a plurality of parts including the register mark, and each part is divided horizontally and vertically. Then, a plurality of pixels are provided, the digital gradation of each pixel is measured, the density histogram is calculated to determine the density centroid level, the density centroid level is compared with the digital gradation of each pixel, and each pixel is Each part is binarized, the image of each part is converted into a binarized feature image, the difference operation between the feature image and the register feature image is performed, and the register mark is detected at the position where the difference operation result is the minimum. Image version The position of the register mark portion, a registration combined method, characterized in that the affine transformation on the register mark position of the reference image version of the detected register mark of the image plate. The present invention also sets the number of pixels in the horizontal and vertical directions of an image having a register mark of a known shape,
The mark is divided into a plurality of pixels, the register mark is binarized for each pixel, the means for converting the register mark into a binarized register characteristic image, and the image version for detecting the register mark are registered in the register.・ Dividing into multiple parts including marks, setting the number of horizontal and vertical pixels in each part, dividing each part into multiple pixels, measuring the digital gradation of each pixel, and calculating the density histogram Determining the density center of gravity level, comparing the density center of gravity level with the digital gradation of each pixel, binarizing each pixel, and converting the image of each part into a binarized characteristic image; A means for performing a difference calculation between the image and the register characteristic image to detect a position on the characteristic image where the difference calculation result is the minimum, and the position is sequentially obtained for each part of the image plane, and the plurality of positions are used as a reference. Image version of multiple A registration alignment apparatus, characterized in that it comprises a means for affine transformation on register marks.

According to the present invention, an image having register marks of known shape is divided in the horizontal and vertical directions, and the register marks are divided into two pixels, for example, "0" and "1". It is digitized and converted into a register feature image. By this, regardless of the shape of the register mark,
The register mark is converted into a register feature image. The image version that should detect the register mark is
It is divided into a plurality of parts including marks, for example, four parts. Each part is further divided into pixels, the digital gradation of each pixel is measured, the density centroid of each part is obtained, the digital gradation of each pixel and the density centroid are compared, and binarized for each pixel. To be done. Since the register mark of each part has a digital gradation larger than the density centroid, for example, "0" and 2
Valued. As a result, the register mark is converted into a characteristic image regardless of the state of the image of the portion and compared with the register characteristic image, so that the position of the register mark is accurately obtained, whereby each image plate with respect to the reference image plate is obtained. Can be easily registered.

[0008]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, the present invention will be described more specifically by way of embodiments, but the present invention is not limited thereto.

An image having a known shape, for example, an image having a register mark shown in FIG. 1 is set with the number of pixels in the horizontal direction and the vertical direction being, for example, 14,8, and each pixel M
Register every (x, y) (x = 0 to x, y = 0 to y)
The white part and the black part of the mark are binarized into, for example, "1" and "0" to create a register characteristic image as shown in FIG. The register feature image does not have to have the same shape as the actual register mark, or may be a part of the image showing the feature of the register mark.

Next, the image plane 1 in which the register mark is to be detected is divided into parts including the register mark 2. Since the register marks 2 are generally provided at the four corners, four parts 1a, 1b, 1 are provided as shown in FIG.
Divide into c and 1d. As shown in FIG. 4, each portion, for example, the portion 1a, has a pixel number i + in the horizontal and vertical directions.
1, j + 1 and each pixel L (i, j) (i = 0 to
i, j = 0 to j), the density is read out in digital gradation of 0 to 255, which is digitized by the color of the image plate,
The density centroid level G is obtained from the density histogram of the relationship between the density and the frequency of the density as shown in FIG. The density centroid level G is compared with the digitized density of each pixel, each pixel is binarized into "0" and "1" by the formula (1) or (2), and the image of each part is characterized. Convert to image.

If L (i, j) ≧ G, L (i, j) = 0 (1) If L (i, j) <G, L (i, j) = 1 (2) Register mark Since No. 2 is darkly printed, a feature image similar to the register feature image shown in FIG. 2 is displayed near the register mark 2. Therefore, the difference calculation shown in Expression (3) is performed between the characteristic image and the register characteristic image, and the position (m, n) at which the difference value becomes the minimum is the register.
Register mark 12 of portion 11a where mark should be detected
Position.

[0012]

[Equation 1]

In addition, this operation is performed by the other parts 1b, 1c, 1d.
Then, the position of the register mark 12 in each part is detected.

The positions of the register marks 2 are detected, for example, at four positions for each color image plate, and the positions of the detected register marks of each image plate are the register marks of the reference image plate. Is subjected to affine transformation, and each image plane is registered.

FIG. 6 is a block diagram of the image processing apparatus 10 according to the present invention. The image with the register mark 2 is, for example, the film 11 or the photographic paper 12 which is color-separated into four colors. Since the density information of these is analog data, it is input to the image input device 13 and sampled. Part 1
In 3-1 sampling is performed, for example, dividing into 512 × 512 pixels, and in the quantizing unit 13-2, 8 bits, that is, 2
It is converted into an image signal digitized to a density value of 56 levels and stored in the frame memory 14 as a signal having an address for each pixel. On the other hand, a pre-digitized image signal, for example, a signal from the optical disc 15, computer graphics (CG) 16 or the like is directly stored in the frame memory 14 without passing through the image input device 13.
The image signal stored in the frame memory 14 may be converted into an analog signal by the digital / analog (D / A) converter 17 if necessary, and then displayed on the monitor display 18 to check the contents.

The image signal stored in the frame memory 14 is supplied to an image creating section for register mark detection (hereinafter simply referred to as an image creating section) 19. The image creating unit 19
A density histogram calculation unit 19-1 for calculating the density detection frequency of the supplied image signal for each pixel and calculating the density histogram of FIG. 5, and a center-of-gravity level calculation unit 19-2 for calculating the center-of-gravity density therefrom. And a binarization processing unit 19-3 for comparing the density level of each pixel with the density of the center of gravity and binarizing the density level according to the expressions (1) and (2), and supplied from the frame memory 14. The image signal is binarized for each pixel, an address is given, and the image signal is given to the central processing unit (CPU) 20 as a characteristic image. The characteristic image signal given to the CPU 20 is a random access memory (RAM) 2
The image file 22 may be stored in the image file 22 such as a floppy disk or a hard disk, or may be displayed on the display (CRT) 23 to check the contents. Image file 22
The characteristic image signal stored in the
It can also be entered in M21.

An image having a register mark of a known shape is processed by the similar image input device 13. Since the register mark is only a pixel which is printed and a pixel which is not printed, the image signal of the frame memory 14 is immediately binarized and the register characteristic image 24 as shown in FIG.
Is obtained, and this is the register RAM (second RAM) 2
5 is stored.

The register / mark detecting section 26 has a RAM 2
The feature image signal stored in No. 1 and the register feature image signal stored in the second RAM 25 are scanned while performing the difference calculation shown in Expression (3). Then, the position (x, y) at which the difference calculation result becomes the minimum is set to the register mark position display unit 27.
Output to

Since the register marks are provided at the four corners of the image plate as shown in FIG. 3, if the positions of the four register marks of each image plate are determined by the above-mentioned method, this will be the reference image. Affine transformation is performed on the position of the register mark of the plate, and each image plate is registered with the reference image plate.

In the registration of the present invention, the position of the register mark is shown in FIG.
It does not prevent opposite registration of 12a and 12d or 12b and 12c.

[0021]

As described above, according to the present invention, the position of the register mark can be accurately detected and registered without depending on the shape of the register mark or the printing condition. It can be performed.

[Brief description of drawings]

FIG. 1 is a diagram showing an example of a register mark.

FIG. 2 is a diagram of a register feature image in which the register mark of FIG. 1 is binarized.

FIG. 3 is a plan view of the image plate 1.

4 is an enlarged plan view of a portion 1a of the image plate 1. FIG.

FIG. 5 is a diagram showing an example of a density histogram.

FIG. 6 is a block diagram showing an example of an image processing apparatus 10 according to the present invention.

[Explanation of symbols]

1 Printing Plate 2 Register Mark Position 10 Image Processing Device 13 Image Input Device 14 Frame Memory 19 Register Mark Detection Image Creating Unit 20 Central Processing Unit (CPU) 21 Random Access Memory (RAM) 22 Register Random Access Memory for Feature Image (Second RAM) 26 register mark detection unit 27 register mark position display unit

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Norihide Kamei 1-23-37 Edobori, Nishi-ku, Osaka City, Osaka Prefecture Sakata Inx Co., Ltd.

Claims (2)

[Claims] 1. An image having a register mark of a known shape is divided into a horizontal direction and a vertical direction to provide a plurality of pixels, each register pixel is binarized, and the register mark is divided into two. The image version that is converted into a binarized register characteristic image and the register mark should be detected is divided into a plurality of parts including the register mark, and each part is divided horizontally and vertically to provide a plurality of pixels. , The digital gradation of each pixel is measured, the density histogram is calculated to determine the density centroid level, the density centroid level is compared with the digital gradation of each pixel, and binarized for each pixel Images of 2
Converted to a valued characteristic image, the difference operation between the characteristic image and the register characteristic image,
The position where the difference calculation result is the minimum is the position of the register mark of each part of the image plate where the register mark is to be detected, and the position of the detected register mark of the image plate is the register mark of the reference image plate. A registration method characterized by affine transformation. 2. The number of pixels in the horizontal and vertical directions of an image having a register mark of a known shape is set, the register mark is divided into a plurality of pixels, and two register marks are provided for each pixel. A means for converting the register mark into a binarized register characteristic image, and an image plate for detecting the register mark are divided into a plurality of parts including the register mark, and the horizontal and vertical directions of each part are divided. Set the number of pixels of each, divide each part into multiple pixels, measure the digital gradation of each pixel, calculate the density histogram to determine the density centroid level, and determine the density centroid level and the digital gradation of each pixel. And a means for converting the image of each part into a binarized feature image by comparing each of the pixels with each other, and performing a difference operation between the feature image and the register feature image. Be the smallest And a means for affine-transforming the plurality of positions with respect to a plurality of register marks of the reference image plate, the position being sequentially obtained for each portion of the image plate. Registration device.