KR102516482B1 - Defect detection device - Google Patents

Abstract

The present invention detects a display module, a camera module that outputs an image of a product produced in a production process, and a specific defective area that matches a set defective product area among a plurality of product areas divided from the image, and detects the specific defective area Provided is a defect detection device including a control module generating defect pattern information for the object and displaying the defect pattern information on the display module.

Description

Defect detection device {Defect detection device}

The present invention relates to a defect detection device, and more particularly, to a defect detection device that facilitates detection of defects in a product produced in a production process based on deep learning.

Deep learning technology, which has been published in conventional hierarchical deep neural network (DNN)-based facial expression recognition technology (IEEE Access 2019, 7) and is used for building a video-based emotion recognition system, is You can read facial expressions. The performance of this technology was confirmed through experiments on four publicly available facial expression recognition datasets (CK+, JAFFE, FERA, and AffectNet).

In a typical deep learning discriminator structure, when learning data is input, it goes through preprocessing and LBP (Local Binary Pattern), and then extracts feature vectors through feature-based CNN.

In addition, a neutral image is generated using auto-encoding from the input data. Emotions can be classified by considering weights after generating geometric feature vectors from emotion pairs based on neutral images and feature vectors using CNN techniques.

On the other hand, when manufacturing a toothbrush, defective inspection performed requires inspection of various parts such as the tip of the bristles, the part where the bristles are planted, and the handle of the toothbrush. In particular, compared to plastic toothbrushes, defects in bamboo toothbrushes are more diverse. It must be determined with high accuracy.

For this reason, the need for automation of toothbrush defect inspection has increased, but currently only some items (raw materials and dimensions of the final product) can be inspected or the inspection success rate (crack on the toothbrush head, hair quality inspection) is low. .

In addition, in the process of post-processing an important part along with discrimination in the inspection process, post-processing automation is also required to reduce the final defect rate by post-processing the identified products to make them good products, and research to solve these problems is being conducted. are doing

An object of the present invention is to provide a defect detection device that can easily detect defects in a product produced in a production process based on deep learning.

The objects of the present invention are not limited to the above-mentioned objects, and other objects and advantages of the present invention not mentioned above can be understood by the following description and will be more clearly understood by the examples of the present invention. It will also be readily apparent that the objects and advantages of the present invention may be realized by means of the instrumentalities and combinations indicated in the claims.

The defect detection apparatus according to the present invention detects a display module, a camera module that outputs an image of a product produced in a production process, and a specific defect area that matches a set product defect area among a plurality of product areas divided from the image, and a control module generating defective pattern information for the specific defective area and displaying the defective pattern information on the display module.

The control module may include a detector for detecting the specific defective area, a classifier for classifying the defective pattern identified in the specific defective area into a set reference defective pattern, and an information output unit for generating the defective pattern information based on the defective pattern. can

The detector may apply the image to a convolutional neural network to divide the image into the plurality of product areas corresponding to parts of the product, and detect the specific defective area matching the defective product area.

The detector may be an R-CNN detector.

The classifier may identify the defect pattern including at least one of the type, location, and color of the defect in the specific defect area, and classify the defect pattern into the reference defect pattern.

When the defect pattern is input from the classifier, the information output unit may display, on the display module, at least one of the type, location, and color of the defect included in the defect pattern and post-processing information on the product. there is.

The post-processing information may be a waste treatment result if the product is a waste product or a recycling treatment result if the product is a renewable product.

The information output unit may discharge the product to a disposal device or a regeneration device according to the post-processing information.

The defect detection apparatus according to the present invention has an advantage of easily extracting defective products by automatically detecting defects of products produced in the production process by comparing them with defective product areas learned through a deep neural network.

On the other hand, the effects of the present invention are not limited to the effects mentioned above, and various effects may be included within a range apparent to those skilled in the art from the contents to be described below.

1 is a control block diagram showing a control configuration of a failure detection device according to the present invention.
2 is an exemplary view showing an example of an image captured by a defect detection device according to the present invention.
3 is an exemplary view showing a defective product in the defect detection device according to the present invention.

Since the present invention can make various changes and have various embodiments, specific embodiments are illustrated in the drawings and described in detail. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all modifications, equivalents, or substitutes included in the spirit and technical scope of the present invention. Like reference numerals have been used for like elements throughout the description of each figure.

Terms such as first, second, A, and B may be used to describe various components, but the components should not be limited by the terms. These terms are only used for the purpose of distinguishing one component from another. For example, a first element may be termed a second element, and similarly, a second element may be termed a first element, without departing from the scope of the present invention. The term and/or includes a combination of a plurality of related recited items or any one of a plurality of related recited items.

It is understood that when an element is referred to as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, but other elements may exist in the middle. It should be. On the other hand, when an element is referred to as “directly connected” or “directly connected” to another element, it should be understood that no other element exists in the middle.

Terms used in this application are only used to describe specific embodiments, and are not intended to limit the present invention. Expressions in the singular number include plural expressions unless the context clearly dictates otherwise. In this application, the terms "include" or "have" are intended to designate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, but one or more other features It should be understood that the presence or addition of numbers, steps, operations, components, parts, or combinations thereof is not precluded.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related art, and unless explicitly defined in this application, it should not be interpreted in an ideal or excessively formal meaning. don't

Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the accompanying drawings.

1 is a control block diagram showing a control configuration of a failure detection device according to the present invention.

Referring to FIG. 1 , the defect detection device 100 may include a display module 110 , a camera module 120 and a control module 130 .

The display module 110 may display the image M photographed by the camera module 120 and the defective pattern information F output from the control module 130 .

The image M may be an image of a product produced in a production process, but is not limited thereto.

In the embodiment, the defect detection device 100 is described as confirming whether a product produced in a production process is a good product or a defective product.

The camera module 120 may capture an image M and output the image to the display module 110 and the control module 130 .

Here, the Carrera module 120 may be an image sensor or a vision sensor, but is not limited thereto.

The control module 130 may include a detector 132 , a classifier 134 and an information outputter 136 .

The detector 132 may divide the image M into a plurality of product areas including products.

Thereafter, the detector 132 may detect a specific defective area that matches the set defective product area among a plurality of product areas.

That is, the detector 132 applies the image M to a convolutional neural network to divide the product into the plurality of product areas corresponding to the parts of the product, and detect the specific defective area matching the defective product area. .

Here, the detector 132 may be an R-CNN detector. The R-CNN detector is an object detection framework that classifies image areas in an image. Instead of using a sliding window to classify each area, it can process only areas that may contain objects, that is, the above-mentioned plurality of product areas. there is.

At this time, the detector 132 follows a transfer learning workflow, and in transfer learning, a neural network trained on a large-scale image collection such as ImageNet can be used as a starting point for a new classification or detection task.

The product defect area is a comparison image area for various defects for each part of the product, and can be pre-trained through a neural network.

The classifier 134 may classify the defective pattern identified in the specific defective area into a set reference defective pattern.

That is, the classifier 134 may identify the defect pattern including at least one of the type, location, and color of the defect in the specific defect area, and classify the defect pattern into the reference defect pattern.

The information output unit 136 may generate the defective pattern information based on the defective pattern.

That is, the information output unit 136 can determine whether the product is a good product or a defective product based on the presence or absence of the defective pattern.

When the defective pattern is input, the information output unit 136 may display on the display module 110 at least one of the type, location, and color of the defective pattern and post-processing information for the product.

The post-processing information may be a waste treatment result if the product is a waste product or a recycling treatment result if the product is a renewable product, but is not limited thereto.

The information output unit 136 may discharge the product to a disposal device or a recycling device according to the post-processing information, but is not limited thereto.

2 is an exemplary view showing an example of an image captured by the defect detection device according to the present invention, and FIG. 3 is an exemplary view showing a defective product in the defect detection device according to the present invention.

If the images shown in FIGS. 2 and 3 show a toothbrush, it will be described that a defect detection device is applied to detect defects in the toothbrush.

2 may be an exemplary view showing the color and stem pattern of a tree of a bamboo toothbrush.

As shown in FIG. 2, in the case of a bamboo toothbrush, the color of the area where defects are to be determined and the focal length of the discrimination area within the discrimination target are different depending on the characteristics of the material and the product, including the color and pattern of the tree, etc. However, since the shape of the implanted bristles shows randomness, it affects not only the identification of implanted hairs but also the determination of other areas, and it may be difficult to perform an overall inspection with a simple vision inspection.

At this time, the defect detection apparatus 100 may divide product areas for each part of the product, such as a toothbrush head and a toothbrush head, and compare the divided product areas with a set defective product area to detect a defective area.

As such, the defect detection device 100 has advantages of cost competitiveness of products and diversification of sales channels.

First, FIG. 3 describes the types and shapes of defects of the toothbrush corresponding to the defect area detected by the defect detection device 100 shown in FIG. 2 .

Figure 3 (a) is an example showing a hair transplant failure. It can be seen that the amount of bristles of the toothbrush is poor, such as a black dot on a part of the head of the toothbrush.

3 (b) is an example of side head breakage, and in the case of a bamboo toothbrush, head breakage may occur.

3(c) is another example of poor hair transplantation, and it can be seen that the length or direction of bristles is not constant.

FIG. 3(d) is an example of incontinence defect, FIG. 3(e) is a line crack defect, and FIG. 3(f) is an example of a crack defect.

As described above, in the defect detection device 100 of the present invention, defects can be classified into hair transplantation, side head breakage, hair transplant defects, incontinence defects, line crack defects, and crack defects. In another modification, defects may be divided into incontinence, line cracks, and cracks as one crack defect, and may be classified into hair transplant amount, side head breakage, hair transplant length, and crack defects.

Features, structures, effects, etc. described in the embodiments above are included in at least one embodiment of the present invention, and are not necessarily limited to only one embodiment. Furthermore, the features, structures, effects, etc. illustrated in each embodiment can be combined or modified with respect to other embodiments by a person having ordinary knowledge in the field to which the embodiments belong. Therefore, contents related to these combinations and variations should be construed as being included in the scope of the present invention.

In addition, although the above has been described with reference to the embodiments, these are only examples and do not limit the present invention, and those skilled in the art to which the present invention belongs can exemplify the above to the extent that does not deviate from the essential characteristics of the present embodiment. It will be appreciated that various modifications and applications that have not been made are possible. For example, each component specifically shown in the embodiment can be modified and implemented. And differences related to these modifications and applications should be construed as being included in the scope of the present invention as defined in the appended claims.

Claims (8)

A defect detection device for detecting defects in a bamboo toothbrush,
display module;
A camera module outputting an image of the bamboo toothbrush produced in the production process; and
The image is divided into a plurality of product areas, a specific defective area matching a set defective product area is detected among the plurality of product areas, and defect pattern information for the specific defective area is generated and displayed on the display module. Including a control module that
The control module,
a detector for dividing the bamboo toothbrush into a plurality of product areas corresponding to bristles and a toothbrush head by applying the image to a convolutional neural network using R-CNN and detecting a specific defective area matching the defective product area;
If the specific defective area is an area included in the bristle, determine the type of defect including defective hair transplantation, side head breakage, incontinence defect, and crack defect in the specific defective area, the location of the defect within the bristle, and the defect a classifier that checks a defective pattern including information on the color of an area to be performed, and classifies the defective pattern into a reference defective pattern; and
When the defective pattern is input from the classifier, based on the defective pattern, information about the type of defect included in the defective pattern, the location of the defect within the bristles, and the color of an area to determine the defect An information output unit generating the defective pattern information;
If the specific defect area is an area included in the toothbrush handle, the classifier determines the type of defects including incontinence defects and crack defects in the specific defect area, the location of defects in the toothbrush handle, and the area to determine the defects. Characterized in that a defective pattern including information about color is identified and the defective pattern is classified into a reference defective pattern.
defect detection device. delete delete delete delete According to claim 1,
The information output device,
display post-processing information on the bamboo toothbrush on the display module;
The post-processing information,
If the bamboo toothbrush is a waste product, it is a waste product, or if it is a renewable product, it is a waste detection device. delete delete

Images