KR102196114B1 - Apparatus and method for inspecting steel products - Google Patents

Abstract

The present invention relates to a steel product defect inspection apparatus and method.
A steel product defect inspection apparatus according to an aspect of the present invention includes: a first inspection means for generating first image information by photographing a surface of a steel product; Second inspection means for generating second image information by applying a magnetic field to the steel product and sensing a leakage magnetic flux; And a defect detection means for detecting surface defects and internal defects of the steel product based on the first image information and the second image information.

Description

{Apparatus and method for inspecting steel products}

The present invention relates to an apparatus and method for inspecting defects in steel products, and more particularly, to an apparatus and method for inspecting defects in steel products by integrating surface defects and internal defects occurring in steel products.

Defects occurring in steel products can be divided into surface defects that appear on the surface and internal defects that exist inside the product (including subsurface defects that exist just below the surface).

In general, for steel products, surface defects occurring in the production process are detected with human eyes or through a surface inspection device equipped with a camera or the like in the process of producing and transferring the steel products.

However, since internal defects cannot be checked with the camera or human eyes, in the case of the prior art, the inspection process for internal defects is omitted, or if necessary, a separate inspection process is provided to move the sensing unit while the steel product is stopped. Internal defects were examined.

However, the internal defect inspection is relatively more complicated than the surface defect inspection, and when inspecting it in a separate inspection process, it takes a lot of manpower, time, and cost, and there is a problem in that the work efficiency is also inferior.

In addition, since the data that can be analyzed is limited even when the internal defect inspection is performed, although defect detection is possible, it is difficult to accurately detect a specific defect.

Korean Patent Publication No. 10-1658700 Korean Patent Publication No. 10-1745764

The present invention was invented to solve the above-described problems, and an object of the present invention is to provide a steel product defect inspection apparatus and method for integrating inspection for surface defects and internal defects on a steel product.

Another object of the present invention is to generate first image information by photographing the surface of a steel product, sense the leakage magnetic flux of the steel product to generate second image information, and synchronize it in units of frames to precisely inspect the defects of the steel product. It is to provide a steel product defect inspection apparatus and method.

Another object of the present invention is to extract defect feature information from the first image information generated by photographing the surface of the steel product and the second image information generated by sensing the leakage magnetic flux of the steel product, and based on a rule-base. It is to provide a steel product defect inspection apparatus and method capable of specifically classifying the types of defects using the defect classification method of.

Another object of the present invention is to use a deep learning-based defect classification method for the first image information generated by photographing the surface of the steel product and the second image information generated by sensing the leakage magnetic flux of the steel product. Thus, it is to provide a steel product defect inspection apparatus and method capable of specifically distinguishing the type of defect.

For the above purposes, a steel product defect inspection apparatus according to an aspect of the present invention includes: a first inspection means for generating first image information by photographing a surface of a steel product; Second inspection means for generating second image information by applying a magnetic field to the steel product and sensing a leakage magnetic flux; And a defect detection means for detecting surface defects and internal defects of the steel product based on the first image information and the second image information.

Preferably, the second image information is generated corresponding to the location and intensity of the leakage magnetic flux in the sensing area having the same size as the photographing area of the first inspection means.

Preferably, the defect detection means performs frame synchronization on one or more frames constituting the first image information and one or more frames constituting the second image information, and the synchronized frame of the first image information And a surface defect and an internal defect of the steel product are detected based on the frame of the second image information.

Preferably, the defect detection means comprises at least one of photographing time information and photographing area information for each frame constituting the first image information, and sensing time information for each frame constituting the second image information, and The frame synchronization is performed based on at least one of the sensing area information.

On the other hand, a method for inspecting a defect in a steel product according to an aspect of the present invention generates first image information by photographing a surface of a steel product, and generates second image information by applying a magnetic field to the steel product and sensing a leakage magnetic flux. Step to do; Performing frame synchronization on one or more frames constituting the first image information and one or more frames constituting the second image information; And detecting surface defects and internal defects of the steel product based on the synchronized frame of the first image information and the frame of the second image information.

Preferably, the generating of the image information comprises generating photographing area information for each frame constituting the first image information, and generating sensing area information for each frame constituting the second image information. Including a process, and the performing of the synchronization comprises: synchronizing the frame of the first image information and the frame of the second image information corresponding to the same area based on the photographing area information and the sensing area information. Include.

Preferably, the generating of the image information comprises generating photographing time information for each frame constituting the first image information, and generating sensing time information for each frame constituting the second image information. Including a process, the step of performing the synchronization, based on the photographing time information and the sensing time information, a difference between the photographing time and the sensing time corresponds to the frame of the first image information and the second It includes a process of synchronizing frames of image information.

According to the present invention, since the first image information is generated by photographing the surface of the steel product, and second image information is generated by sensing the leakage magnetic flux of the steel product, it is synchronized in a frame unit to inspect the defect of the steel product, It has an effect that can be performed by integrating surface defect inspection and internal defect inspection for steel products.

In particular, according to the present invention, a first area corresponding to the same area of the steel product is generated after generating first image information by photographing a first area and sensing the second area to generate second image information. Since the surface defect inspection and the internal defect inspection are performed in real time by matching the frame of the image information and the frame of the second image information, a separate internal defect inspection process is not required, so manpower, time, and cost can be greatly reduced. , Accordingly, it has an effect of increasing work efficiency.

In addition, according to the present invention, a rule-based defect classification method is primarily applied based on the first image information generated by photographing the surface of the steel product and the second image information generated by sensing the leakage magnetic flux of the steel product. And it has the effect of detecting the defects of steel products more precisely by applying the defect classification method based on deep learning secondarily.

1 is a block diagram of a steel product defect inspection apparatus according to an embodiment of the present invention.
2 is a configuration diagram of a first inspection means according to an embodiment of the present invention.
3 is a block diagram of a second inspection means according to an embodiment of the present invention.
4 is a diagram showing a method of generating image information for a steel product being transferred by a first inspection means and a second inspection means according to an embodiment of the present invention.
5 is a block diagram of a defect detection means according to an embodiment of the present invention.
6 is a configuration diagram of a defect classification means according to an embodiment of the present invention.
7 is a configuration diagram of an inspection setting means according to an embodiment of the present invention.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings and preferred embodiments. For reference, detailed descriptions of known functions and configurations that may unnecessarily obscure the subject matter of the present invention will be omitted in the following description.

1 is a block diagram of a steel product defect inspection apparatus according to an embodiment of the present invention.

1, a steel product defect inspection apparatus 100 according to an embodiment of the present invention includes a first inspection means 110, a second inspection means 120, a defect detection means 130, a defect classification means ( 140), inspection setting means 150, defect display means 160, and the like.

The first inspection means 110 is provided in a path through which steel products are transported and performs a function of inspecting surface defects of steel products.

In this regard, Figure 2 shows the configuration of the first inspection means 110 according to an embodiment of the present invention.

Referring to FIG. 2, a first inspection means 110 according to an embodiment of the present invention includes an illumination unit 111, an image capture unit 112, a first control unit 113, and the like.

The lighting unit 111 is equipped with one or more lights to emit light to the surface of the steel product, and the image photographing unit 112 is equipped with one or more cameras to photograph the surface of the steel product to generate first image information (data). do. In addition, the first control unit 113 is a lighting unit to generate the first image information in a preset frame standard (eg, resolution, frame rate) for a preset location (area) at a preset time with respect to the surface of the steel product. (111), the image photographing unit 112, and the like are controlled. To this end, the first control unit 113 receives a synchronization signal (eg, a trigger pulse of an encoder) and receives other components of the steel product defect inspection apparatus 100 (in particular, the second inspection means ( It is synchronized with the second control unit 124) of 120).

The second inspection means 120 is provided in a path through which the steel product is transported, and is preferably provided adjacent to the first inspection means 110 to perform a function of inspecting surface defects and internal defects of the steel product.

In this regard, Figure 3 shows the configuration of the second inspection means 120 according to an embodiment of the present invention.

3, the second inspection means 120 according to an embodiment of the present invention includes a magnetic field generating unit 121, a sensor unit 122, an imaging processing unit 123, a second control unit 124, etc. do.

The magnetic field generator 121 generates a magnetic field from a steel product, and the sensor unit 122 is provided with one or more magnetic sensors to detect the leakage magnetic flux of the steel product. In addition, the imaging processing unit 123 generates second image information (data) imaged according to the position of each magnetic sensor of the sensor unit 122 and the intensity of the detected leakage magnetic flux, and the second control unit 124 The magnetic field generating unit 121, the sensor unit 122, the imaging processing unit 123, and the like are controlled so that second image information can be generated in a preset frame standard for a preset position for a steel product at a preset time. To this end, the second control unit 124 similarly receives a synchronization signal (eg, trigger pulse of the encoder) and receives other components of the steel product defect inspection apparatus 100 (in particular, the first control unit of the first inspection unit 110 ). (113)).

Hereinafter, with reference to FIG. 4, a description will be given of a method in which the first inspection means and the second inspection means according to an embodiment of the present invention generate image information for a steel product being transferred and synchronize it.

Referring to FIG. 4, the image photographing unit 112 of the first inspection means 110 and the sensor unit 122 of the second inspection means 120 are provided at a predetermined distance on the path through which the steel products are transferred, Each image information is generated for the same steel product.

Specifically, the image photographing unit 112 of the first inspection means 110 is provided on the path through which the steel product is transported and photographs a preset first region (shooting region) in a predetermined time unit, and is composed of one or more frames. 1 Create (acquire) video information. And, the first control unit 113 of the first inspection means 110, when the first image information is generated by the image capturing unit 112, shooting time information and/or shooting area information for each frame of the first image information Create

Similarly, the sensor unit 122 of the second inspection means 120 is provided on the path through which the steel product is transported to detect the strength of the leakage magnetic flux with respect to a preset second area (sensing area), and the second inspection means ( The imaging processing unit 123 of 120) generates second image information (consisting of one or more frames) corresponding to the location and intensity of the leakage magnetic flux. And, the second control unit 124 of the second inspection means 120, when generating the second image information in the imaging processing unit 123, the sensing time information and/or sensing region information for each frame of the second image information. Generate.

For reference, the image information (frame) corresponding to the location and intensity of the leakage magnetic flux, when the sensor unit 122 includes a plurality of magnetic sensors in a two-dimensional sensing structure (planar sensing structure), the position of each magnetic sensor and It may be generated using a two-dimensional interpolation method, etc. based on the intensity of the leakage magnetic flux detected at the corresponding location (this is expressed in color in the image), and the sensor unit 122 senses one or more magnetic sensors in one dimension. In the case of having a structure (linear sensing structure), one-dimensional linear sensing data for a moving steel product can be obtained continuously or in a very short period by a distance corresponding to the second area, and then generated using a two-dimensional interpolation method. have.

On the other hand, the first area image of the first inspection means 110 and the second area image of the second inspection means 120 generated as described above are a position (area) separated by a predetermined distance in the longitudinal direction (transport direction) of the steel product. The image of the steel product defect inspection apparatus 100 according to an embodiment of the present invention is a frame for the first image information of the first inspection means 110 and the second image information of the second inspection means 120 Synchronization is performed.

Specifically, when the first inspection means 110 generates image information (first image information) of the first region, the first image information (consisting of one or more frames) according to a preset frame standard (e.g., resolution, frame rate) ) Is generated, and at this time, capturing time information and/or capturing area information for each frame of the first image information are generated together. In addition, the first inspection means 110 transmits the first image information composed of one or more frames and the photographing time information and/or the photographing area information matching each frame of the first image information to the defect detection means 130 together. do.

Similarly, when the second inspection means 120 generates the image information (second image information) of the second area, the second image information is the same frame standard as the first inspection means 110 (eg, resolution, frame rate). (Consisting of one or more frames) is generated, and at this time, sensing time information and/or sensing region information for each frame of the second image information are generated together. In addition, the second inspection means 120 transmits second image information composed of one or more frames and sensing time information and/or sensing region information matching each frame of the second image information to the defect detection means 130 together. do.

Then, the defect detection means 130 uses the photographing time information and the sensing time information and/or the photographing region information and the sensing region information to form a frame and a second image of the first image information corresponding to the same location (area) of the steel product. Frame synchronization is performed by matching frames of information.

For reference, the photographing time information and the sensing time information are synchronized with each other. When frame synchronization is performed using the photographing time information and the sensing time information, the first region (the photographing region of the first image information) and the second region (the second region 2 Calculate the inspection delay time (the difference between the shooting time and the sensing time for the same area of the steel product) based on the separation distance between the image information sensing area and the feed speed of the steel product, and perform frame synchronization based on this. I can. In addition, the photographing area information and the sensing area information can be obtained by calculating the transfer distance based on the transfer speed and transfer time based on the leading edge (start position) of the steel product. When frame synchronization is performed by using, frame synchronization may be performed by matching the frame of the first image information and the frame of the second image information corresponding to the same area (location).

Referring back to FIG. 1, the defect detection means 130 receives first image information, photographing time information, photographing area information, etc. from the first inspection means 110, and receives a second image from the second inspection means 120. By receiving information, sensing time information, and sensing area information, surface defects and internal defects of steel products are detected.

In this regard, FIG. 5 shows the configuration of the defect detection means 130 according to an embodiment of the present invention.

5, the defect detection means 130 according to an embodiment of the present invention includes a frame synchronization unit 131, a defect detection unit 132, a defect analysis unit 133, a defect image storage unit 134, and the like. Include.

The frame synchronization unit 131 synchronizes the first image information transmitted from the first inspection unit 110 and the second image information transmitted from the second inspection unit 120 on a frame-by-frame basis. As described above with reference to FIG. 4, the frame synchronization unit 131 uses the shooting time information matched with each frame of the first image information and sensing time information matched with each frame of the second image information, and/ Or, using the photographing area information matched with each frame of the first image information and the sensing area information matched with each frame of the second image information, the frame of the first image information corresponding to the same location (area) of the steel product and Frame synchronization is performed by matching frames of the second image information.

The defect detection unit 132 applies each defect detection criterion (defect determination criterion) to the frame of the first image information and the frame of the second image information synchronized by the frame synchronization unit 131 on a frame-by-frame basis. Is detected. For example, the defect detection unit 132 determines whether it is a normal state or a defect state by applying a first defect detection criterion to a frame of first image information, and applies a second defect detection criterion to a frame of second image information. After determining whether it is a normal state or a defective state, if either of the synchronized frame of the first image information and the frame of the second image information is in a defective state, it is determined that there is a defect, and a defect of the steel product is detected.

The defect analysis unit 133 analyzes the defect of the steel product detected by the defect detection unit 132 and extracts defect characteristic information. For example, the defect analysis unit 133 analyzes the frame of the first image information and the frame of the second image information synchronized with each other determined to be defective by the defect detection unit 132 to determine the location, size, shape, and shape of the defect. Defect feature information such as shadow value, distribution map, and shadow ratio is extracted.

In addition, the defective image storage unit 134 stores the frame of the first image information and the frame of the second image information synchronized with each other determined to be defective by the defect detection unit 132.

Referring back to FIG. 1, the defect classification means 140 accurately determines which defect is a specific defect using at least one defect classification method for the defect of the steel product detected by the defect detection means 130.

In this regard, Fig. 6 shows the configuration of the defect classification means 140 according to an embodiment of the present invention.

6, the defect classification means 140 according to an embodiment of the present invention includes a first defect classification unit 141, a second defect classification unit 142, a post-processing unit 143, and a defect result storage unit ( 144) and the like.

The first defect classification unit 141 uses a rule-based defect classification method based on the defect characteristic information extracted from the defect analysis unit 133 of the defect detection unit 130 Classify. For example, the first defect classification unit 141 applies defect classification criteria for the location, size, shape, shadow value, distribution, and shadow ratio of the defect to the defect characteristic information to determine which defect corresponds to the defect of the steel product. .

The second defect classification unit 142 is based on deep learning for the frame of the first image information and the frame of the second image information synchronized with each other determined to be defective by the defect detection unit of the defect detection unit 130 Classify defects in steel products using the defect classification method of For example, the second defect classification unit 142 refers to the defect learning data learned through defect classification learning for the frame of the first image information and the frame of the second image information synchronized to each other, Determine if it is applicable.

According to a preferred embodiment of the present invention, the first defect classification unit 141 primarily determines the defect type of the steel product based on the defect characteristic information, and the second defect classification unit 141 determines the deep learning learning data. By secondly determining the types of defects in steel products with reference, it is possible to more accurately determine defects in steel products.

The post-processing unit 143 separates or in addition to the rule-based defect classification method of the first defect classification unit 141 and the deep learning-based defect classification method of the second defect classification unit 142 The defect classification method set by is applied to determine which defect is the defect of the steel product.

In addition, the defect result storage unit 144 stores the defect result of the steel product determined in detail through the first defect classification unit 141, the second defect classification unit 142, the post-processing unit 143, and the like.

Referring back to FIG. 1, the inspection setting means 150 provides a user interface so that the user can directly set a defect detection function, a defect classification function, a post-processing function, and the like, and the defect detection means 130 according to the user's setting. Adjust the defect detection function details of the defect classification means 140 and the defect classification function details and post-processing function details.

In this regard, FIG. 7 shows the configuration of the inspection setting means 150 according to an embodiment of the present invention.

Referring to FIG. 7, the inspection setting means 150 according to an embodiment of the present invention includes a defect detection setting unit 151, a defect classification setting unit 152, a post processing setting unit 153, and the like.

The defect detection setting unit 151 provides a user interface for directly setting the defect detection function to the user, and provides details on a frame synchronization method, a defect detection standard, and a defect feature information extraction method according to the user's setting. Transfer to 130 to set the defect detection function of the defect detection means (130).

The defect classification setting unit 152 provides a user interface for directly setting the defect classification function to the user, and determines the details of the rule-based defect classification method and the deep learning-based defect classification method according to the user's setting. Transmitted to the classification means 140, the defect classification function of the defect classification means 140 is set.

The post-processing setting unit 153 provides a user interface for directly setting the post-processing function to the user, and transmits details of the defect classification method to the defect classification unit 140 according to the user's setting to the defect classification unit ( 140) post-processing function.

Referring back to FIG. 1, the defect display unit 160 notifies the user of the defect of the steel product finally classified by the defect classification unit 140 through a display, LED, alarm sound, etc., and the user When a defect of a product is inquired, the defect result stored in the defect classification means 140 is retrieved and displayed.

Although the present invention has been described in detail with reference to preferred embodiments so far, those skilled in the art to which the present invention pertains can be implemented in various other specific forms without changing the technical spirit or essential features of the present invention. The embodiments described above are illustrative in all respects and should be understood as non-limiting.

And, the scope of the present invention is specified by the claims to be described later rather than the detailed description, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts are included in the scope of the present invention. Should be interpreted as.

Claims (14)

A first image information for a photographing area defined by the width and length of the steel product, and photographing time information and photographing region information for each frame constituting the first image information, by photographing the surface of the steel product 1 inspection means;
Sensing time information and sensing for each frame constituting the second image information and the second image information on the sensing area defined by the width and length of the steel product by applying a magnetic field to the steel product and sensing the leakage magnetic flux Second inspection means for generating area information;
In at least one of information on a photographing time and information on each frame constituting the first image information, and information on at least one of information on a sensing time and information on each frame constituting the second image information A frame synchronization unit that synchronizes the frame of the first image information and the frame of the second image information corresponding to the same region based on the frame unit, and the synchronized frame of the first image information and the second A defect detection unit configured to detect defects of the steel product based on a frame of image information, and a defect analysis unit configured to extract defect characteristic information by analyzing defects of the steel product detected by the defect detection unit. Defect detection means for detecting and analyzing the data; And
A first defect classification unit that primarily determines the type of defect of the steel product using a rule-based defect classification method based on the defect characteristic information, and the synchronized first image information Secondly determining the type of defect of the steel product using a defect classification method based on deep learning by referring to defect learning data learned through defect classification learning for the frame and the frame of the second image information. And a defect classifying means for classifying whether the defect of the steel product is a surface defect or an internal defect, comprising a second defect classification unit. The method of claim 1,
The second image information is generated according to the position and intensity of the leakage magnetic flux in the sensing area having the same size as the photographing area of the first inspection means. delete delete The method according to claim 1 or 2,
The first inspection means,
An image photographing unit that photographs the surface of the steel product and generates the first image information composed of one or more frames; And
And a first control unit that generates at least one of information on a photographing time and information on a photographing area for each frame of the first image information. The method of claim 5,
The second inspection means,
A magnetic field generator for generating a magnetic field with the steel product;
A sensor unit sensing the leakage magnetic flux of the steel product;
An imaging processor configured to generate the second image information composed of one or more frames corresponding to the position and intensity of the leakage magnetic flux sensed by the sensor unit; And
And a second control unit that generates at least one of sensing time information and sensing region information for each frame of the second image information. delete delete delete By photographing the surface of the steel product, first image information for a photographing region defined by the width and length of the steel product, and photographing time information and photographing region information for each frame constituting the first image information are generated, and the Second image information on the sensing area defined by the width and length of the steel product by applying a magnetic field to the steel product and sensing the leakage magnetic flux, and sensing time information and sensing area information for each frame constituting the second image information Generating a;
In at least one of information on a photographing time and information on each frame constituting the first image information, and information on at least one of information on a sensing time and information on each frame constituting the second image information Synchronizing the frame of the first image information and the frame of the second image information corresponding to the same region on a frame-by-frame basis;
Detecting a defect in the steel product based on the synchronized frame of the first image information and the frame of the second image information;
Analyzing the detected defect of the steel product and extracting defect characteristic information; And
Based on the defect characteristic information, the defect type of the steel product is primarily determined using a rule-base based defect classification method, and the synchronized frame of the first image information and the second image Defects of the steel product are secondarily determined by using a deep learning-based defect classification method with reference to the defect learning data learned through defect classification learning for the frame of information. And classifying whether the defect is a surface defect or an internal defect. delete The method of claim 10,
The photographing area information and the sensing area information are generated by calculating a conveying distance based on a conveying speed and a conveying time based on a leading edge of the steel product. The method of claim 10,
The synchronizing on a per-frame basis may include a frame of the first image information and a frame of the second image information in which a difference between a capturing time and a sensing time corresponds to an inspection delay time based on the capturing time information and the sensing time information Steel product defect inspection method comprising the process of synchronizing. The method of claim 13,
The inspection delay time is calculated based on a separation distance between a photographing area of the first image information and a sensing area of the second image information and a transfer speed of the steel product.

Images