JP2900600B2 - Pattern inspection method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pattern inspection method used to determine whether a pattern displayed on a surface of an object by printing or the like is a predetermined pattern.

<Related Art> For example, in a production line that handles products of various kinds of small lot production, it is necessary to check whether or not a product group of a predetermined kind is flowing correctly on the production line. Usually, characters such as a lot number indicating a product type and graphics (hereinafter, collectively referred to as "patterns") are displayed on the surface of each product.
The worker inspects these patterns to check whether different types of products are mixed.

However, this kind of inspection method has drawbacks that not only human labor is required, but also inspection errors are likely to occur, and the inspection efficiency is low.

Therefore, it has been proposed to automate the discrimination of a pattern using a visual sensor.

In this pattern inspection method, a pattern displayed on a product flowing through a line is imaged using an imaging device, and the image is determined to determine whether the pattern to be inspected is a predetermined pattern.

<Problems to be Solved by the Invention> However, when handling products manufactured in a large variety of small lots, it is necessary for an operator to specify a pattern to be determined each time using a keyboard or the like, which is troublesome. In addition, there is a problem that a designation error is likely to occur because any one of a variety of products is designated, and when there is a designation mistake, a pattern to be inspected cannot be properly determined, leading to confusion.

The present invention has been made in view of the above problems, and has as its object to provide a pattern inspection method which simplifies inspection work and completely prevents occurrence of designation errors.

<Means for Solving the Problems> In the present invention, the following teaching step and inspection step are performed to inspect whether or not the pattern to be inspected is a predetermined pattern. That is, in the teaching step, after obtaining the center of gravity of an image obtained by imaging the model of the pattern to be inspected, the center of a predetermined window is aligned with the position of the center of gravity, and each window is gradually expanded or reduced while the window is gradually expanded or reduced. The area of the image included in the window of the step is measured and stored as reference data. In the inspection step, after the center of gravity of the image obtained by imaging the pattern to be inspected is obtained, the center of the window is aligned with the position of the center of gravity of the image, and the window is stepped at the same ratio as the teaching step. The pattern to be inspected is determined by measuring the area of the image included in the window at each stage while expanding or reducing the size, and comparing the measured data with the corresponding reference data.

<Operation> To determine whether or not the pattern to be inspected is a predetermined pattern, a model of the pattern to be inspected is imaged prior to the inspection step to generate reference data. Is compared with the reference data to determine the pattern to be inspected, so that the inspection work is simplified and the occurrence of a designation error can be prevented as compared with the conventional method of specifying the pattern to be determined using a keyboard or the like. In addition, since the center of gravity is obtained for the image of the pattern to be inspected and the window is aligned with the position of the center of gravity to obtain reference data and measurement data, even if there is a pattern misalignment, it is not affected by it and proper collation is possible. It is.

<Embodiment> FIG. 1 shows the appearance of a pattern inspection apparatus used for carrying out the present invention.

The pattern inspection apparatus of the illustrated example is for discriminating a product type of each product group of a multi-product small lot production flowing on a production line by a pattern such as a character displayed on an upper surface.

In FIG. 1, reference numeral 1 denotes a transport belt that transports the inspected product 2 to an inspection position P. A position sensor (not shown) is provided at the inspection position P to detect the arrival of the inspected product 2 at the inspection position P. The conveyor belt 1 is driven by a drive mechanism in the direction of arrow a, and stops driving each time the inspected product 2 reaches the inspection position P.
To the inspection position P.

At a position adjacent to the inspection position P, a table 3 having substantially the same height as the conveyor belt 1 is arranged. A model 2 'of the inspection object is placed on the table 3 and a teaching process described later is performed. .

Above the inspection position P or the table 3, the imaging device 4
Lighting device (not shown) such as a circular light source is arranged,
The imaging device 4 is connected to the image processing device 5.

A CCD camera or the like is used as the imaging device 4, and the objective lens 6 is positioned and fixed above the inspection position P or above the table 3 with the object lens 6 facing downward.

In this embodiment, the teaching process is carried out by placing the model 2 'of the product to be inspected on the table 3. However, the table 3 is not always necessary. The teaching step may be performed at the position 2 '.

The image processing device 5 includes a microcomputer and the like,
An image obtained by imaging the pattern 7 of the inspection object 2 by the imaging device 4 is subjected to a process such as binarization to determine whether the pattern 7 is a predetermined pattern.

FIG. 2 shows the principle of the pattern inspection method performed by the image processing apparatus 5.

In the figure, reference numeral 8 denotes a binary image of the character pattern "A", where the hatched portion is an image region of black pixels, and the other portions are image regions of white pixels. G indicates the position of the center of gravity of the binary image 8, and a first window W1 having a circular shape whose center is aligned with the position of the center of gravity G, and the first window W1 are stepped with respect to the binary image 8. , The second to eighth windows W2 to W8 which form an annular shape extending in the entire circumferential direction are set.

FIG. 3 shows the relationship between the first window W1 to the third window W3 as an example. The first window W1 has a circular shape with a diameter d, and the second window W2 has an inner diameter d. The third window W3 has an annular shape having an inner diameter of 2d and an outer diameter of 3d. The fourth and subsequent windows W4 to W8 also have the same relationship as described above, and are not illustrated here. In these windows W1 to W3, the hatched portion indicates the measurement target area, and the window setting is performed on the binary image 8 while performing the above-described extended operation, and the number of black pixels included in each measurement target area is determined by the feature amount. Is counted as

The window is not limited to a circular or annular shape, but may be a rectangular or rectangular annular shape as shown in FIG. In the case of the embodiment shown in FIG. 4, a rectangular first window whose center is aligned with the center of gravity G of the binary image 8
The second and subsequent windows W2 to W6 are formed by gradually expanding W1 in all directions.

In each of the above examples, the other window is generated by expanding the smallest window W1, but the other windows may be set by gradually reducing the size based on the largest windows W8 and W6. .

FIG. 5 shows an example of setting a window for a pattern 9 consisting of a character string.

In the case of the illustrated example, the first window W1 having a rectangular shape whose center is aligned with the center position G of the pattern 9 is gradually expanded to both sides so that each of the second and subsequent windows can be formed.
W2 to W5 are formed.

FIG. 6 shows, as an example, the relationship between the first window W1 to the third window W3. The first window W1 has a rectangular shape having a width D, and the second window W2 has an inner width D. , A 3D parallel rectangle with an outer width of 3D, the third window W3
Has a parallel rectangular shape with an inner width of 2D and an outer width of 5D. The fourth and subsequent windows W4 to W5 also have the same relationship as described above, and are not shown here. In these windows W1 to W3, the shaded portion means the measurement target area, similarly to the above example.

7 and 8 show a specific procedure of the pattern inspection method based on the above principle, that is, a control procedure by the image processing apparatus 5. FIG.

FIG. 7 shows a teaching process performed prior to the inspection. At the start time, the inspection object is placed as a model 2 'on the table 3 and the next step 1 ("ST
1), an optimum window shape (for example, the window shape shown in FIGS. 2 and 3) for inspecting the pattern 7 of the inspection object 2 is selected. After setting the counter in the image processing device 5 to “1” in step 2, the image pickup device 4 images the model 2 ′ of the inspection object in the next step 3, and the image signal is input to the image processing device 5. (Step 3). The image processing device 5 binarizes the input image to generate a binary image 8, stores the binary image in the image memory, and measures the center of gravity G of the binary image 8 (step 4).

In the next step 5, it is determined whether or not the count value of the counter i has exceeded the window set number n (N = 8 in FIGS. 2 and 3). In this case, this determination is “NO”. If so, the process proceeds to step 6, where the center is positioned at the center of gravity G of the binary image 8, and a first window W1 is set. In the next step 7, the number of black pixels A (1) of the binary image 8 included in the measurement target area of the first window W1 is measured as a feature amount and stored in the memory.

Next, after the counter i is incremented in step 8, a second window W2 is set for the binary image 8, the same measurement of the number of black pixels A (2) is performed, and the measurement data is stored in the memory. (Steps 6 and 7).

When all the windows W1 to W8 are sequentially set in this way and the measurement of the number of black pixels A (1) to A (n) is completed, the determination in step 5 becomes "YES", the teaching process is completed, and The process proceeds to the inspection process shown in FIG.

When the conveyor belt is driven at the start time in FIG. 8, the inspected products 2 are sequentially guided to the inspection position P. First, in step 1, the counter i is set to "1", and in the next step 2, the pattern 7 of the inspection object 2 is imaged by the imaging device 4. In the following steps, the same procedure as the teaching process, that is, the center of gravity measurement (step 3), the windows W1 to
W8 setting (step 5) and black pixel measurement (step 6) are executed, and the number of black pixels B (1) to B (1) in each window W1 to W8.
B (8) is measured.

When the measurement process is completed, step 4 becomes “YES” and the process proceeds to step 8, where the measurement data for each window, that is, A (0) and B (0), A (1) and B (1), A
(2) is compared with B (2), ‥‥, and whether or not the corresponding measurement data matches, that is, the difference | A (0) −B (0) |, | A ( 1) -B (1) |, | A (2)
It is determined whether each of −B (2) |, ‥‥ is equal to or less than a predetermined threshold value (step 9). If the determination in step 9 is “YES”, an inspection result indicating that the pattern 7 of the inspection object 2 is a pattern to be determined is output (step 1).
0). On the other hand, if the determination in step 9 is "NO", an inspection result indicating that the pattern 7 of the inspection object is not a pattern to be determined is output (step 11).

In step 12, it is determined whether or not there is a subsequent inspection object 2. If the determination is "NO", the inspection process is completed. If "YES", the process returns to step 1 and returns to step 1. The same procedure as described above is executed for the product 2 to be inspected.

<Effects of the Invention> As described above, in the present invention, in order to determine whether a pattern to be inspected is a predetermined pattern, an image of a model of the pattern to be inspected is generated prior to an inspection step, and reference data is generated. In the inspection process, the pattern to be inspected is determined by comparing the measured data of the pattern to be inspected with the reference data, so that the inspection work can be performed in comparison with the conventional method of specifying the pattern to be determined using a keyboard or the like. Simplification is possible, and the occurrence of designation errors can be prevented. In addition, since the center of gravity is obtained for the image of the pattern to be inspected, and the window is aligned with the position of the center of gravity to obtain the reference data and the measured data, even if there is a pattern displacement, it is not affected by the displacement, and the proper position is obtained. There is a remarkable effect of achieving the object of the invention, such as enabling collation.

[Brief description of the drawings]

FIG. 1 is a perspective view showing an external appearance of a pattern inspection apparatus used for carrying out the present invention, FIG. 2 is an explanatory view showing the principle of the pattern inspection method of the present invention, FIG. 3 is an explanatory view showing the shape of a window, 4 and 5 are explanatory diagrams showing examples of setting other windows, FIG. 6 is an explanatory diagram showing the shape of the window in FIG. 5, and FIGS. 7 and 8 are pattern inspection methods according to the present invention. 6 is a flowchart showing the procedure of FIG. 2… Inspection object, 4… Imaging device 5 …… Image processing device W1-W8… Window

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-64-83102 (JP, A) JP-A-60-143704 (JP, A) Full-fledged Sho-55-134661 (JP, U) (58) Field (Int.Cl. ⁶ , DB name) G01B 11/00-11/30 G06T 5/20

Claims (1)

(57) [Claims] 1. A method for inspecting whether or not a pattern to be inspected is a predetermined pattern, comprising: obtaining a center of gravity of an image obtained by imaging a model of the pattern to be inspected; A teaching step of aligning the center of the window, gradually expanding or reducing the window, measuring and storing the area of the image included in the window at each stage as reference data, and capturing the pattern to be inspected. After calculating the center of gravity of the image obtained
The center of the window is aligned with the center of gravity of the image, and the area of the image included in the window at each stage is measured while gradually expanding or reducing the window at the same ratio as the teaching process. An inspection step of comparing each measurement data with the corresponding reference data to determine a pattern to be inspected.