JP7601327B2 - Inspection Equipment - Google Patents

Description

The present invention relates to an inspection device that can inspect the surface condition of an object, for example, an inspection device that can inspect the printing condition of multiple types of mixed product packages.

Generally, product packaging is printed with an image of the product, as well as text and symbols that explain the product.

However, in such product packaging, multiple types of product packaging may be printed for the same type of content. For example, when considering product packaging for hair dye, there are many products that correspond to different hair colors, and the product packaging is printed with different colors and designs accordingly. In addition, when considering product packaging for food products, there are many products with different flavors for the same type of product, such as seasonal or regional limited products, and the corresponding product packaging is printed to have the respective image.

When inspecting the printing condition of inspection objects such as multiple types of product packages, reference data is generally stored for each inspection object, and the printing condition of the inspection object is inspected by comparing data based on an image of the inspection object taken by a camera with the reference data (e.g., Patent Document 1).

JP 2010-160039 A

However, when inspecting multiple types of objects in this way by comparing them with their respective reference data, the following problems arise:

For example, when inspecting the printing condition of multiple types of inspection objects, each type must be separated and the data based on the image obtained from the inspection object must be compared with the reference data corresponding to the type of inspection object. Furthermore, when changing the type of inspection object, the reference data that serves as the inspection standard must be changed in advance, and when inspecting dozens or more types of inspection objects, this task becomes extremely tedious.

On the other hand, if these inspection objects are placed on a conveyor in a mixed state and photographed by a camera to compare with the reference data, they must be inspected against all types of reference data, which takes time. Furthermore, if the inspection results in an item being determined to be defective, it is not clear which reference data was used to determine the defect, making it impossible to determine whether the defect was determined based on a different type of reference data, or whether the item was truly defective based on the relevant type of reference data.

Therefore, the present invention has been made with a focus on the above problem, and aims to provide an inspection device that can quickly inspect objects even when multiple types of objects are mixed together.

That is, in order to solve the above-mentioned problems, the present invention provides a system comprising a camera which captures an image of a transported product package, an alignment means which aligns the image captured by the camera using a specified shape and extracts an image of the same shape which has been aligned, a first storage means which stores in advance feature data indicating the features of the product package for each of the multiple types of product packages having the same shape, a second storage means which stores reference data of an entire image for each of the multiple types of product packages having the same shape, a feature data extraction means which acquires the feature data of the product package from the image having the same shape extracted by the alignment means, and an inspection means which reads out from the first storage means a type of product package which matches the feature data acquired by the feature data extraction means, reads out the reference data of the product package of that type from the second storage means, and compares it with data based on the entire image of the product package captured by the camera to inspect the condition of the product package .

According to the present invention, it becomes possible to inspect the printing condition of multiple types of product packages of the same shape, for example, those that differ in color.

1 is a schematic diagram of an inspection device according to an embodiment of the present invention; Functional block diagram of the inspection device in the same configuration. A memory map diagram for storing feature data and reference data in the same form. FIG. 13 shows characteristic data of an object to be inspected in the same form. FIG. 13 shows a flow for registering feature data in the same form. A flowchart showing the inspection process in the embodiment. A flowchart showing an inspection process according to another embodiment.

One embodiment of the present invention will be described below with reference to the drawings.

The inspection device 1 in this embodiment is capable of inspecting the printing state of multiple types (e.g., types with different colors) of products (e.g., hair dye) enclosed in the same shaped product packages 9a. Characteristically, among the product characters, figures, and symbols printed on the product packages 9a, some feature data that can be distinguished from other product packages 9a is registered and stored in the first storage means 5a, and the feature data is linked to reference data for the overall inspection of the product packages 9a and stored in the second storage means 5b. The feature data is extracted from the image captured by the camera 3 to determine which product package 9a it is, and the reference data linked to the feature data is read out and compared with data based on the image of the product package 9a for inspection. The inspection device 1 in this embodiment is described in detail below.

First, the camera 3 that captures an image of the inspection object 9 is composed of a line sensor, a CCD sensor, or the like, and as shown in FIG. 1, captures a color image from the inspection object 9 transported by the conveyor 2. This camera 3 is set up so that it can capture an image of the entire inspection object 9, and a lighting device is provided at a predetermined position corresponding to this camera 3. Since the inspection state changes depending on the angle of the light, this lighting device is set up so that it can irradiate light from multiple angles, and it may also be possible to adjust the illuminance, color, etc.

The image captured by the camera 3 is output to the alignment means 4 (see the block diagram in Figure 2), where it is aligned by processing such as rotation and translation. At this time, since the objects to be inspected 9 have the same shape, alignment is performed using a specified shape. The image is then oriented in a certain direction so that it can be compared with reference data, etc. Note that, although the alignment means 4 is used to align the images here, if an object to be inspected 9 with a completely different shape or if overlapping objects to be inspected 9 are introduced, alignment cannot be performed, and an error message may be displayed.

Next, feature data is extracted from the image aligned by the alignment means 4 in this manner. This feature data has features that can be distinguished from other types of inspection objects 9 (product packages 9a), and from within the product packages 9a, data on the product image and parts of the product's explanatory text and symbols is used. For example, parts of text, figures, symbols, etc. that are less affected by rotation are used as such feature data, and the image and position of the characteristic part is stored as feature data in the first storage means 5a by the operator.

To go into more detail about this feature data, as shown in FIG. 4, for the first type of product package 9a, an image of the apex area of the letter "A" written thereon is stored as feature data, and for the second type of product package 9a, an image of the center area of the letter "B" written thereon is stored as feature data. For the third type of product package 9a, an image of the upper opening area of the letter "C" written thereon is stored as feature data, and for the fourth type of product package 9a, an image of the upper left corner area of the letter "D" written thereon is stored as feature data in the first storage means 5a.

On the other hand, the second storage means 5b stores reference data corresponding to the feature data stored in the first storage means 5a, as shown in FIG. 4. This reference data is data for comparison with data based on the overall image of the product package 9a to be inspected, and is generated from image data from a product package 9a that is determined to be a good product, data created when the product package 9a is designed, etc. Specifically, information such as the outline shape patterns of characters, figures, symbols, etc., and colors is used as the reference data, which is linked to the feature data of that type of product package 9a and stored in the second storage means 5b. Note that when the feature data and the reference data are stored in pairs in the storage means in this way, if the number of product packages 9a becomes very large, there is a possibility that the feature data will coincide with the feature data of a product package 9a that has already been registered. Therefore, in such a case, the following method can be adopted.

As a first method, as shown in FIG. 5, when an operator sets feature data (step R1), the feature data is compared with feature data already stored in the first storage means 5a (step R2) to determine whether there is reference data with a matching rate higher than a predetermined threshold. If the matching rate is lower than the predetermined threshold, the feature data is stored in the first storage means 5a (step R3), and if the matching rate is higher than the predetermined threshold, the feature data is displayed so that it cannot be registered (step R4), and the operator is prompted to register other feature data.

Alternatively, as another method, the feature data registered by the operator may be stored as is in the first storage means 5a regardless of whether the feature data matches, and inspection may be performed using all reference data stored in association with that feature data. Then, if the result of inspection using all reference data that matches the feature data is that it is determined to be appropriate, the product is passed through inspection as a "good product." However, when inspection is performed using multiple pieces of inspection data in this way, for example, it may be possible to display a message indicating that the inspection was performed using multiple pieces of inspection data, or a message indicating which type of inspection object 9's reference data was used.

The feature data extraction means 6 extracts feature data from the image of the inspection object 9 captured by the camera 3. When extracting this feature data, an image of an area stored in the first storage means 5a as an area of feature data is extracted in advance, and data based on the image is compared with the feature data to extract the type of feature data that has a higher matching rate than a predetermined threshold. Then, the discrimination means 61 discriminates the type of inspection object 9 that is similar to the feature data, and reads out the reference data that is linked to and stored with this type of feature data from the second storage means 5b.

The inspection means 7 compares the reference data thus read out with data based on the overall image acquired from the inspection object 9 to determine whether the inspection object 9 is good or bad. Specifically, a method of extracting data such as characters, figures, and symbols from the acquired overall image and inspecting whether this data matches the reference data, or a method of extracting RGB luminance values for a specific area and inspecting color tone by comparing them with the RGB luminance values of the reference data can be used. Of course, inspection can be performed by other methods, and inspection can be made for faint printing, peeling of paper, etc.

The output means 8 notifies the operator of the results of this inspection. Alternatively, if the system is equipped with a function for automatically separating defective products, it outputs information indicating that the products are "defective" to the separating device, and the defective products are collected separately.

Next, a flowchart for inspecting the inspection target 9 using the inspection device 1 configured in this manner will be described with reference to FIG. 6. As a premise, it is assumed that the printing state of multiple types of product packages 9a for the same product is to be inspected, and the product packages 9a are randomly placed on the conveyor with the same side facing up, and then photographed by the camera 3 for inspection.

First, when a product package 9a is placed on the conveyor 2 and flows down, the camera 3 captures an overall image of the product package 9a (step S1). The image is then output to the alignment means 4, where it is aligned by rotation, translation, etc., and corrected to face a fixed direction (step S2).

Next, from the image thus corrected by positioning, the feature data extraction means 6 extracts a portion of the image feature data (step S3). At this time, an image of an area previously stored as feature data in the first storage means 5a is acquired, and it is determined whether the data based on the image matches the feature data stored in the first storage means 5a (step S4). Then, the type of product package 9a with the feature data that has the highest match rate, which is higher than a predetermined value, is determined, and the reference data in the second storage means 5b linked to that feature data is read (step S5). On the other hand, as a result of comparing with all the feature data (step S6), if there is no feature data with a match rate higher than the predetermined threshold, it is possible that the product package 9a is placed face down or that the product packages 9a are being transported overlapping each other, and an error is output (step S7).

When the reference data is read out in this manner (step S5), the data based on the overall image acquired in step S1 is compared with the reference data, and the condition of the inspection object 9 is inspected (step S8). If the matching rate exceeds a predetermined threshold, the inspection object is deemed to be a "good product" (step S9). Conversely, if the matching rate is equal to or lower than the predetermined threshold, the inspection object is deemed to be a "defective product" (step S10), and this is output via the output means 8.

Next, as a second embodiment, we will explain the inspection method when the same feature data is unconditionally registered, using the flowchart in Figure 7.

As in FIG. 6, when a product package 9a is placed on the conveyor and conveyed, an image of the entire product package 9a is acquired by the camera 3 (step T1), and the image is aligned by rotation, translation, etc., and corrected to face in a fixed direction (step T2).

Next, from the image aligned in this way, the feature data extraction means 6 is used to extract a portion of the feature data of the image (step T3). At this time, an image of an area previously stored as feature data in the first storage means 5a is acquired, and it is determined whether the data based on the image matches the feature data stored in the first storage means 5a (step T4). Then, all types of feature data with a matching rate higher than a predetermined threshold are extracted, and all reference data linked to the feature data are read from the second storage means 5b (step T5). At this time, if there are multiple extracted feature data, multiple reference data corresponding to the feature data are read. On the other hand, as a result of comparing with all the feature data (step T6), if there is no feature data with a matching rate higher than the predetermined threshold, it is possible that the product package 9a is placed face down or that the product packages 9a are being transported overlapping each other, and an error is output (step T7).

When multiple reference data are read out in this way (step T5), the data based on the overall image acquired in step T1 is compared with each of the reference data to inspect the state of the inspection target 9 (step T8). For reference data whose matching rate exceeds a predetermined threshold, a message indicating that the data is a "good product" along with its type is output (step T9), and for all other cases, a message indicating that the data is a "defective product" is output (step T10).

As described above, according to the embodiment, the system includes the camera 3 which captures an image of the transported product package 9a, the alignment means 4 which aligns the image captured by the camera 3 using a specified shape and extracts an image of the same shape which has been aligned, the first storage means 5a which stores in advance feature data indicating the features of the product package 9a for each of the multiple types of product packages 9a having the same shape, the second storage means 5b which stores reference data of an entire image for each of the multiple types of product packages 9a having the same shape, the feature data extraction means 6 which captures the feature data of the product package 9a from the image of the same shape extracted by the alignment means 4, and the inspection means 7 which reads out from the first storage means 5a the type of product package 9a which matches the feature data captured by the feature data extraction means 6, reads out from the second storage means 5b the reference data for that type of product package 9a, and compares it with data based on the entire image of the product package 9a captured by the camera 3 to inspect the condition of the product package 9a. Therefore, it becomes possible to inspect the printing conditions of multiple types of product packages having the same shape, for example, by differences in color.

The present invention is not limited to the above embodiment and can be implemented in various ways.

For example, in the above embodiment, the case of inspecting the printing condition of the product package 9a has been described as an example, but the present invention can also be applied to inspecting the printing condition of books, flyers, etc., or inspecting the shape and surface condition of the product itself. However, when inspecting these inspection objects 9, it is preferable that they all have the same shape.

In the above embodiment, a portion of the feature data is used to identify the type of inspection object 9 and the reference data associated with it is extracted, but when inspecting a huge variety of inspection objects 9, multiple types of feature data may be set. Alternatively, before extracting the feature data, the product group may be roughly selected and sorted into certain product groups based on, for example, the average brightness value of the acquired overall image, and feature data may be extracted from the product groups in a narrowed-down manner.

Furthermore, in the above embodiment, the feature data is stored in the first storage means 5a and the reference data is stored in the second storage means 5b, but these may be stored in the same storage means.

1: Inspection device 3: Camera
4. Alignment means
5a, 5b...Memory means 6...Feature data extraction means 61...Discrimination means 7...Inspection means 9...Inspection object (9a product package)

Claims (1)

A camera that captures images of the delivered product packages;
A registration means for performing registration using a specified shape from the images acquired by the camera and extracting images having the same shape that have been registered;
a first storage means for storing in advance characteristic data indicating characteristics of the product package for each of the plurality of types of product packages having the same shape;
a second storage means for storing reference data of an entire image for each of the plurality of types of product packages having the same shape;
a feature data extraction means for acquiring feature data of a product package from the images having the same shape extracted by the positioning means;
an inspection means for reading out from the first storage means a type of product package that matches the characteristic data acquired by the characteristic data extraction means, reading out reference data for the product package of that type from the second storage means, and comparing the data with data based on the overall image of the product package acquired by the camera to inspect the state of the product package;
An inspection device comprising:

Images