JP7408488B2 - Coke oven photography equipment, coke oven inspection equipment, and image processing equipment - Google Patents

Description

The present invention relates to a coke oven photographing device, a coke oven inspection device, an image processing device, and a photographing device for industrial equipment.

Traditionally, in order to inspect for damage to the walls of the carbonization chamber of a coke oven, instead of inspecting the inside of the coke oven directly, inspectors inserted a camera equipped with heat protection into the oven and photographed the oven wall. , it has been proposed to have the inspector observe the image. The entrance to a coke oven is narrow, and the interior of the oven is quite long and narrow, so the camera usually only captures a small portion of the oven wall at a time. Therefore, by sequentially taking pictures while moving the camera in the depth direction and stitching these images together, an entire furnace wall image is obtained.

JP2011-126988A

The present inventors studied the above coke oven wall observation method and came to recognize the following problems. The environment inside a coke oven is harsh for photographing; for example, when a coke oven is in operation, a large amount of dust from the coke may fly around inside the oven. Because visibility is poor, it may be difficult to take clear photographs of the furnace wall. Further, the inside of the furnace is heated to a high temperature of, for example, 1000° C. or higher. The color and intensity of light emitted from the furnace wall due to high temperatures changes depending on the temperature, so the appearance of the furnace wall depends on the fluctuations in temperature inside the furnace that occur during coke oven operation and the temperature distribution on the furnace wall. can be affected by temperature differences. This can be alleviated by using high-resolution cameras, but then the imaging systems tend to be expensive and bulky. Due to these circumstances, in reality, it is not always easy to take good images of the furnace wall for damage inspection.

In addition to photographing the inside of a coke oven, when photographing the surface to be inspected of other industrial equipment, if the photographing conditions or the state of the surface to be inspected change during the photographing process, it is necessary to obtain better images for inspection. It is desirable to make this possible.

One exemplary objective of certain aspects of the present invention is to generate images of inspected surfaces suitable for damage inspection in coke ovens or other industrial equipment.

According to an aspect of the present invention, the coke oven photographing device includes a camera that photographs the oven wall surface as a perspective image while moving inside the coke oven, and a camera that captures at least one of the oven wall surfaces based on a series of oven interior images sequentially output from the camera. and an image processing unit that generates a furnace wall surface image including a portion of the furnace wall surface. The image processing unit cuts out the same region of interest on the furnace wall from at least two images of the series of furnace interior images and the generated furnace wall surface images, and generates an inspection image of the region of interest based on the cut out partial images. .

According to an aspect of the present invention, a coke oven inspection device includes the coke oven photographing device according to the above aspect, and a processing device that receives an inspection image as input and outputs information regarding damage to the oven wall surface through image recognition. .

According to an aspect of the present invention, the image processing device receives a series of furnace interior images sequentially taken as perspective images of the furnace wall surface while moving the camera within the coke oven, and based on the series of furnace interior images. The apparatus includes an image processing unit that generates a furnace wall surface image including at least a portion of the furnace wall surface. The image processing unit cuts out the same region of interest on the furnace wall from at least two images of the series of furnace interior images and the generated furnace wall surface images, and generates an inspection image of the region of interest based on the cut out partial images. .

According to an aspect of the present invention, an imaging device for industrial equipment includes a camera that takes a perspective image of a surface to be inspected of the industrial equipment while moving inside or around the industrial equipment, and a series of camera images sequentially output from the camera. an image processing unit that generates an image of the surface to be inspected including at least a portion of the surface to be inspected based on the image processing unit. The image processing unit cuts out the same region of interest on the surface to be inspected from each of at least two of the series of camera images and the generated image of the surface to be inspected, and generates an inspection image of the region of interest based on the cut out partial images. do.

Note that arbitrary combinations of the above-mentioned constituent elements and mutual substitution of constituent elements and expressions of the present invention among methods, devices, systems, etc. are also effective as aspects of the present invention.

According to the present invention, it is possible to generate an image of a surface to be inspected suitable for damage inspection in a coke oven or other industrial equipment.

FIG. 1 is a side view schematically showing an in-furnace observation device according to an embodiment. 2 is a diagram schematically showing the inside of a coke oven as seen from the observation device shown in FIG. 1. FIG. FIG. 1 is a block diagram of a coke oven photographing device according to an embodiment. FIGS. 4(a) and 4(b) are diagrams showing examples of in-furnace images. FIG. 3 is a diagram showing an example of a front view image of a furnace wall surface. FIG. 1 is a block diagram of a coke oven photographing device according to an embodiment. FIGS. 7A and 7B are diagrams showing exemplary configurations of the image partial region selection section. FIGS. 8A and 8B are diagrams showing examples of user interface screens displayed on the display.

DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments for carrying out the present invention will be described in detail with reference to the drawings. In the description and drawings, the same or equivalent components, members, and processes are denoted by the same reference numerals, and overlapping explanations will be omitted as appropriate. The scales and shapes of the parts shown in the figures are set for convenience to facilitate explanation, and should not be interpreted in a limited manner unless otherwise stated. The embodiments are illustrative and do not limit the scope of the present invention. All features and combinations thereof described in the embodiments are not necessarily essential to the invention.

FIG. 1 is a side view schematically showing an in-furnace observation device according to an embodiment. The furnace interior observation device shown in FIG. 1 is used to observe the inside of a coke oven carbonization chamber (hereinafter also simply referred to as a coke oven). Moreover, FIG. 2 is a diagram schematically showing the inside of the coke oven when viewed from the observation device shown in FIG. 1.

The coke oven 10 is a narrow oven with a pair of brick oven walls 11 facing each other. The oven wall 11 extends, for example, more than ten meters in the depth direction from the oven inlet 12 on one side of the coke oven 10 to the oven outlet 13 on the opposite side. The interval between the furnace walls 11 is, for example, several tens of centimeters. The height from the bottom 15 of the coke oven 10 to the ceiling 14 is, for example, several meters.

An extrusion device 20 is provided that repeatedly moves back and forth within the coke oven 10. On the outward journey, the extrusion device 20 is inserted into the coke oven 10 from the oven inlet 12 and pushes out the coke C produced in the coke oven 10 to the oven outlet 13. On the return trip, the extrusion device 20 returns inside the coke oven 10 from the oven outlet 13 to the oven inlet 12. The extrusion device 20 includes a push plate 21 and a beam 22. The beam 22 connects the push plate 21 to a drive device (not shown), and the drive device moves the push plate 21 from the oven inlet 12 to the oven outlet 13 of the coke oven 10. It is possible to move freely up to Since the push plate 21 has the same shape as the cross section of the coke oven 10, the coke C can be pushed out by moving the push plate 21.

The camera 30 continuously photographs the oven wall 11 while moving in the depth direction inside the coke oven 10 together with the extrusion device 20. Camera 30 may be a still camera or a video camera. The camera 30 is attached to a support stand installed on the back of the push plate 21 or behind the push plate 21. The camera 30 is mounted so as to face in a direction opposite to the direction in which the coke C is extruded by the extrusion device 20 (to the right in FIG. 1) so that the view is not obstructed by the push plate 21 or the coke C. The camera 30 is installed with the depth direction as the front, and can photograph the furnace walls 11 on both the left and right sides. The furnace wall 11 is photographed as a perspective image. In order to protect the camera 30 from the high-temperature environment (for example, 1000° C. or higher) inside the coke oven 10, the camera 30 is provided with heat countermeasures such as being housed in a heat-resistant housing or a cooling box, for example.

FIG. 3 is a block diagram of the coke oven photographing apparatus 100 according to the embodiment. The coke oven photographing device 100 includes an image processing section 110 in addition to the camera 30. Further, the arithmetic processing device 210 constitutes a coke oven inspection device 200 together with the coke oven photographing device 100. The image processing unit 110 and the arithmetic processing unit 210 are implemented by a combination of a processor (hardware) such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or a microcomputer, and a software program executed by the processor (hardware). be able to.

The camera 30 is configured to output a series of sequentially captured in-furnace images to the image processing unit 110. For example, the camera 30 may be connected to the image processing unit 110 via a communication cable when taking pictures while moving inside the coke oven, and may output images of the inside of the oven to the image processing unit 110 in real time while taking pictures. Alternatively, the camera 30 may store the in-furnace image in a built-in memory, and output the in-furnace image to the image processing unit 110 outside the coke oven after capturing the image. In this case, the communication cable does not need to be connected to the camera 30 during shooting.

The coke oven photographing device 100 is configured to generate camera position information indicating the position at which the camera 30 has photographed each of a series of inside images. For example, a camera position sensor may be provided that measures the camera position in the depth direction within the coke oven while the camera 30 is taking pictures. As the camera position sensor, for example, an encoder installed in the extrusion device to measure the position of the extrusion device in the depth direction within the coke oven may be used. Camera position information may be input to the image processing unit 110 from the camera position sensor, and the image processing unit 110 may associate each in-furnace image with its photographing position.

An image data storage device that functions as an image database that can be accessed by the image processing section 110 may be provided inside or outside the image processing section 110, and a series of in-furnace images sequentially output from the camera 30 are stored in the image database. May be accumulated. The in-furnace images accumulated in the image database may be processed by the image processing unit 110, for example, in real time or at a desired timing (off-line). In addition to the in-furnace images captured by the camera 30, the image database may also store front-view images of the furnace wall output from the image processing unit 110 or other images.

The image processing unit 110 generates a furnace wall surface image including at least a portion of the furnace wall surface based on a series of furnace interior images sequentially outputted from the camera 30. The image processing unit 110 may generate a front view image of at least a portion of the furnace wall surface as the furnace wall surface image. The front-view image of the furnace wall surface is generated by converting the furnace interior image, which is a perspective image, into a front-view image through viewpoint conversion processing, and joining the obtained front-view images. Such processing for generating a front-view image of the furnace wall surface can be performed using existing methods as appropriate, and will not be described in detail here.

Note that the furnace wall surface image is not limited to a front-view image. The furnace wall surface image may be a perspective image, and the image processing unit 110 may select and output one or more furnace interior images from a series of furnace interior images sequentially output from the camera 30. The selected furnace interior image can be regarded as the above-mentioned furnace wall surface image.

FIGS. 4(a) and 4(b) show examples of in-furnace images. FIG. 5 shows an example of a front view image of the furnace wall surface. The in-furnace image 32a shown in FIG. 4(a) is taken on the outward path of the reciprocating movement of the extrusion device 20 shown in FIG. 1, and the in-furnace image 32b shown in FIG. 4(b) is taken during the same reciprocating movement. On the return trip, the photograph is taken at the same in-furnace position as the in-furnace image 32a. The front view image 34 shown in FIG. 5 is generated based on a series of in-furnace images including in-furnace images 32a and 32b.

The image processing unit 110 cuts out the same target region 36 on the furnace wall surface from each of at least two images out of the series of furnace interior images. The region of interest 36 may be based on a user's selection. For example, the user selects a region of interest as the region of interest 36 in one of the series of furnace images (for example, one of the furnace images 32a and 32b). Such selection of the region of interest 36 may be input to the image processing unit 110 through a user interface provided in the coke oven imaging device 100. The image processing unit 110 may identify the same target region 36 for one or more other in-furnace images (for example, the other in-furnace image among the in-furnace images 32a and 32b) based on the camera position information. . In this way, the image processing unit 110 cuts out a partial image including the same region of interest 36 from each of at least two images (for example, the in-furnace images 32a and 32b) of the series of in-furnace images (FIGS. 3A and 3 (indicated by a rectangle in (b)). Note that the selection of the region of interest 36 and the cutting out of partial images are not limited to being performed using a perspective image of the furnace interior image, but may also be performed using a front view image of the furnace wall surface that is generated based on the furnace interior image. good.

The image processing unit 110 generates an inspection image of the region of interest 36 based on the cut out partial image. When the partial image is cut out from the in-furnace image, the inspection image of the region of interest 36 may be a perspective image like the in-furnace image. The image processing unit 110 may convert the inspection image, which is a perspective image, into a front view image. Further, when the partial image is cut out from the front view image, the inspection image of the region of interest 36 becomes the front view image.

The inspection image of the region of interest 36 generated by the image processing unit 110 may be presented to the user as an image for visual inspection, for example by displaying it on a display. Alternatively, the inspection image of the region of interest 36 generated by the image processing unit 110 may be input to the arithmetic processing device 210, as shown in FIG.

The arithmetic processing unit 210 inputs the inspection image (for example, the front-view image 34) and outputs information regarding damage to the furnace wall surface through image recognition. For example, the processing unit 210 uses deep learning to identify the type of damage area selected in advance, such as cracks, chips, holes, and carbon deposits on the furnace wall surface based on the input inspection image. It is constructed using machine learning. Alternatively, if applicable, the processing unit 210 may be based on pattern matching between the input inspection image and a large number of sample image data prepared in advance for different parts of the furnace wall surface and different damages, respectively. .

Therefore, the coke oven inspection apparatus 200 detects a damaged area on the oven wall surface by image recognition, such as area division or object detection, based on the input inspection image, and presents the position of the damaged area to the user. The coke oven inspection device 200 may detect not only the position of the damaged area but also the type of the damaged area, such as cracks, chips, holes, carbon deposits, etc., by image recognition, and present the type of the damaged area to the user. The coke oven inspection device 200 may present to the user a healthy area (i.e., an area without damage) on the oven wall surface together with or in place of the damaged area.

According to the coke oven photographing device 100 according to the embodiment, the image processing unit 110 cuts out the same target region 36 on the oven wall surface from at least two images from each of the series of oven interior images sequentially output from the camera 30, A plurality of partial images including the region of interest 36 are acquired. The image processing unit 110 generates an inspection image of the region of interest 36 based on the plurality of cut out partial images. In this way, a plurality of partial images of the same region of interest 36 taken at different timings can be prepared. Even if any one of the plurality of partial images is difficult to observe due to the effects of dust or furnace temperature, any other partial image may be an image that is less affected by such effects. By using a plurality of partial images of the same region of interest 36, it is expected that the possibility of obtaining an inspection image suitable for observing the furnace wall surface will be increased compared to the case based on only one image. This allows for more accurate damage inspection.

At least two of the series of in-furnace images may include an in-furnace image taken on a return trip during reciprocating movement of the extrusion device, or a front-view image generated based on the in-furnace image. On the outward journey, the extrusion device pushes out the coke, which tends to generate a lot of dust inside the coke oven. On the other hand, on the return trip, since the coke has already been discharged from inside the coke oven, there is less dust than on the outward trip, and the camera 30 can take a clearer image of the inside of the oven. Therefore, it is possible to select a partial image (part of interest 36) suitable for observation of the furnace wall surface based on the in-furnace image taken on the return trip.

At least two images that cut out a plurality of partial images of the same region of interest 36 are also conceivable. For example, the at least two images may include a pair of in-oven images taken at the same in-oven location on the outbound and return trips of one round trip of the camera within the coke oven. The at least two images may include a plurality of in-oven images taken at different times during either the outward or return trip of a single round trip of the camera within the coke oven. The at least two images may include a plurality of in-oven images taken at the same in-oven location over multiple reciprocating movements of the camera within the coke oven. The at least two images may include the generated front-view images. The at least two images are two different front-view images generated based on a series of inside-oven images during one or more reciprocating movements of the camera within the coke oven (e.g., front-view images obtained from each outbound and return trip). visual images). Even in this case, it is expected that the possibility of obtaining an inspection image suitable for observing the furnace wall surface will increase.

Further, from the viewpoint of machine learning, it is preferable that at least two images obtained by cutting out a plurality of partial images of the same region of interest 36 do not include images with different properties. Therefore, at least two images may both be in-furnace images. Alternatively, at least two images may both be front-view images.

FIG. 6 is a block diagram of a coke oven photographing apparatus 100 according to an embodiment. An example of a specific configuration of the coke oven photographing apparatus 100 will be described with reference to FIG. 6. The coke oven photographing device 100 includes an input image photographing section 111, an image partial region acquisition section 112, an image partial region selection section 113, an image sharpening processing section 114, an image recognition processing section 115, and an output image display section 116.

The input image photographing unit 111 is an image acquisition means and corresponds to the camera 30 in FIG. 3 . The input image capturing unit 111 may include a camera, means for moving the camera, means for acquiring the position or movement distance of the camera, means for controlling the attitude of the camera, and the like. The image photographed by the input image photographing section 111 is input to the image partial area acquisition section 112 together with position information.

The image partial area acquisition unit 112 and the image partial area selection unit 113 can be associated with the image processing unit 110 shown in FIG. The image sharpening processing section 114 and the image recognition processing section 115 can be associated with the arithmetic processing device 210 shown in FIG.

The image partial area acquisition unit 112 extracts a plurality of partial images including the same region of interest on the furnace wall surface from the image group input from the input image capturing unit 111. The input image group may include a series of in-furnace images sequentially output from the camera, and a furnace wall image generated based on the series of in-furnace images, such as a front-view image.

The image partial area selection unit 113 selects partial images effective for damage inspection of the furnace wall surface from the plurality of partial images extracted by the image partial area acquisition unit 112, and selects an inspection image of the region of interest from the selected effective partial images. generate. Alternatively, the image partial area selection unit 113 synthesizes effective images that are effective for damage inspection of the furnace wall surface from the plurality of partial images extracted by the image partial area acquisition unit 112, and generates an inspection image of the region of interest from the effective images. do.

FIGS. 7A and 7B show exemplary configurations of the image partial area selection unit 113. As shown in FIG. 7(a), the image partial region selection unit 113 performs machine learning identification to estimate whether the input furnace interior image or its partial image is effective for damage inspection of the furnace wall surface. A container 117 is provided. The image partial area selection unit 113 sequentially inputs the cut out partial images 40 to the discriminator 117, selects partial images effective for damage inspection of the furnace wall surface based on the output 42 from the discriminator 117, and selects the selected partial images. An inspection image of the region of interest is generated from the valid partial images.

The output 42 of the discriminator 117 may indicate in a binary manner whether or not the input partial image 40 is effective for damage inspection (for example, the output 42 may indicate whether the input partial image 40 is effective for damage inspection). If it is not valid, it may take the value 1; if it is not valid, it may take the value 0). Alternatively, the output 42 of the discriminator 117 may indicate the probability that the input partial image 40 is valid for damage inspection. The image partial region selection unit 113 may acquire the output 42 for each of the plurality of partial images 40, and select the partial image 40 with the highest probability of being effective for damage inspection as the partial image effective for damage inspection. Alternatively, a partial image 40 whose output 42 exceeds a certain threshold may be selected as a partial image effective for damage inspection.

The classifier 117 may be configured with a convolutional neural network (CNN). For learning of the classifier 117, pairs of input image groups and manual evaluation results of validity/invalidity for each input image may be used as training data. Note that the classifier 117 may be constructed using a non-deep learning method such as a support vector machine.

As shown in FIG. 7(b), the image partial area selection unit 113 performs machine learning to synthesize an effective image that is effective for damage inspection of the furnace wall surface from the input furnace interior image group or its partial image group. An image generator 118 is provided. The image partial area selection unit 113 inputs the cut out partial image 40 to the image generator 118, synthesizes the valid image 44, and generates an inspection image of the region of interest from the valid image 44.

Image generator 118 may be configured with a convolutional neural network (CNN). Image generator 118 may be configured to synthesize an image from a predetermined number of images. For the learning of the image generator 118, a pair of a predetermined number of input images and a correct image manually evaluated as effective among the input images may be used as training data.

Referring again to FIG. 6, the image sharpening processing unit 114 applies at least one of super-resolution processing, denoising processing, and deblurring processing to the image input from the image partial region selection unit 113. Performs image sharpening processing including: Super-resolution processing enlarges an image to a predetermined size without deteriorating it (or while suppressing image deterioration). The denoising process removes or reduces noise in the input image, such as camera noise and salt-and-pepper noise. The deblurring process corrects defocus, blur of moving objects, and the like.

Such image sharpening processing can be implemented in a variety of ways. For example, for super-resolution processing, a simple interpolation method, an example-based method, or a super-resolution model obtained by machine learning can be employed. These super-resolution models are, for example, "R. Timofte, V.D. Smet, and L.V.Gool. A+: Adjusted Anchored Neighborhood Regression for Fast Super-Resolution. ACCV, 2014.", "C. Dong, C.C. Loy, K. He , and X. Tang. Image Super-Resolution Using Deep Convolutional Networks. TPAMI, 2016.” Regarding denoising/debulus processing, for example, smoothing processing using a Gaussian filter and methods using machine learning can be considered. BMVC, 2015. Ru.

Note that the image sharpening processing unit 114 is an option and may be omitted. The output image from the image partial region selection section 113 may be input to the image recognition processing section 115.

The image recognition processing unit 115 performs damage detection on the image input from the image sharpening processing unit 114 (or image partial area selection unit 113) based on image recognition, such as area division and object detection, and outputs the results. do. The image recognition processing unit 115 may be configured with a convolutional neural network (CNN), and the damage detection processing may be performed using, for example, "J. Shotton, J. Winn, C. Rother, and A. Criminisi, TextonBoost: Joint Appearance," Shape and Context Modeling for Multi-Class Object Recognition and Segmentation. ECCV, 2006.” and “J. Long, E. Shelhamer, and T. Darrell, Fully Convolutional Networks for Semantic Segmentation. CVPR, 2015.” may be adopted.

The output image display section 116 outputs the damage inspection results obtained by the image recognition processing section 115. The output image display unit 116 may include a display and visually present the test results to the user. The output image display unit 116 may store data indicating the test results in a storage medium.

FIGS. 8A and 8B show examples of user interface screens displayed on the display 122. As shown in FIG. 8(a), a reference image 300 (for example, any of the above-mentioned furnace interior image, furnace wall surface image, front-view image, or inspection image) and an inspection result image 302 showing the damaged area 204 are They may be displayed side by side. As shown in FIG. 8(b), an inspection result image 302 in which the damaged area 204 and the reference image are superimposed may be displayed. An enlargement frame 304 for enlarging an area selected by the user, such as the damaged area 204, may be displayed on the display 122 together with the inspection image 302. In addition, the user may have a function to select a target area (damage area) found on the screen, and a function to display target area information such as the position, size, and type of the target area.

The present invention has been described above based on examples. It will be understood by those skilled in the art that the present invention is not limited to the above embodiments, and that various design changes and modifications are possible, and that such modifications also fall within the scope of the present invention. By the way. Various features described in connection with one embodiment are also applicable to other embodiments. A new embodiment resulting from a combination has the effects of each of the combined embodiments.

In the above-described embodiment, the case where the photographing device and the inspection device are applied to a coke oven is described as an example, but the present invention is not limited to this, and can be applied to various industrial equipment such as industrial machinery, plants, and social infrastructure. may be applied to. In one embodiment, an industrial equipment imaging device includes a camera that captures a perspective image of a surface to be inspected of the industrial equipment while moving within or around the industrial equipment, and a series of camera images sequentially output from the camera. An image processing unit that generates an image of the surface to be inspected of at least a portion of the surface to be inspected may be provided. The surface to be inspected may be an inner wall surface or an outer wall surface of industrial equipment. The camera may be mounted on a flying vehicle such as a drone to take pictures of the exterior wall surface of industrial equipment. The image processing unit cuts out the same region of interest on the surface to be inspected from each of at least two of the series of camera images and the generated image of the surface to be inspected, and generates an inspection image of the region of interest based on the cut out partial images. You may. Further, an inspection device for industrial equipment may include the above-described photographing device and a processing device that receives an inspection image as input and outputs information regarding damage to the surface to be inspected through image recognition.

Although the present invention has been described using specific words based on the embodiments, the embodiments merely illustrate one aspect of the principles and applications of the present invention, and the embodiments do not include the claims. Many modifications and changes in arrangement are possible without departing from the spirit of the invention as defined in scope.

30 camera, 32a, 32b furnace image, 34 front view image, 36 region of interest, 40 partial image, 42 output, 44 effective image, 100 coke oven photographing device, 110 image processing unit, 117 discriminator, 118 image generator, 200 Coke oven inspection equipment, 210 Arithmetic processing unit.

Claims (9)

A camera that takes perspective images of the oven wall while moving inside the coke oven;
an image processing unit that generates a furnace wall surface image including at least a portion of the furnace wall surface based on a series of furnace interior images sequentially output from the camera;
The image processing unit cuts out the same region of interest on the furnace wall surface from at least two images taken at different timings from among the series of furnace interior images and the generated furnace wall surface images, and based on the cut out partial images. generate an inspection image of the region of interest ,
The image processing unit includes a machine learning discriminator that estimates whether the input furnace interior image or a partial image thereof is effective for damage inspection of the furnace wall surface,
The image processing unit sequentially inputs the cut out partial images to the discriminator, selects a partial image effective for damage inspection of the furnace wall surface based on the output from the discriminator, and selects a partial image effective for damage inspection of the furnace wall surface. A coke oven photographing device characterized in that an inspection image of the region of interest is generated from a partial image . A camera that takes perspective images of the oven wall while moving inside the coke oven;
an image processing unit that generates a furnace wall surface image including at least a portion of the furnace wall surface based on a series of furnace interior images sequentially output from the camera;
The image processing unit cuts out the same region of interest on the furnace wall surface from at least two images taken at different timings from among the series of furnace interior images and the generated furnace wall surface images, and based on the cut out partial images. generate an inspection image of the region of interest,
The image processing unit includes a machine learning image generator that synthesizes an effective image that is effective for damage inspection of the furnace wall surface from an input group of in-furnace images or a group of partial images thereof,
The coke oven is characterized in that the image processing unit inputs the cut out partial images to the image generator, synthesizes the effective images, and generates an inspection image of the region of interest from the effective images. Photography equipment. The camera is mounted on an extrusion device that repeatedly moves back and forth within the coke oven,
3. The at least two images include an in-furnace image taken during a return trip during reciprocating movement of the extrusion device, or an in-furnace image generated based on the in-furnace image. coke oven photographic equipment. The at least two images are
a pair of in-furnace images taken at the same in-furnace position on the outbound and return trips during one reciprocating movement of the camera in the coke oven;
a plurality of furnace interior images taken at different times during either the outward or return trip of one reciprocating movement of the camera within the coke oven, and
4. Any one of claims 1 to 3, comprising at least one of a plurality of furnace interior images taken at the same furnace position over a plurality of reciprocating movements of the camera within the coke oven. The coke oven photographing device described. The image processing unit generates a front view image of at least a portion of the furnace wall surface as the furnace wall surface image,
The coke oven photographing apparatus according to any one of claims 1 to 4 , wherein the at least two images include the front view image. The at least two images include two different front-view images generated based on the series of oven interior images during one or more reciprocating movements of the camera within the coke oven. A coke oven photographing device according to claim 5 . A coke oven photographing device according to any one of claims 1 to 6 ,
A coke oven inspection device comprising: an arithmetic processing unit that receives the inspection image as input and outputs information regarding damage to the oven wall surface through image recognition. Receive a series of furnace interior images taken sequentially by the camera as perspective images of the furnace wall surface while moving the camera within the coke oven, and based on the series of furnace interior images, a furnace wall surface image including at least a portion of the furnace wall surface. Equipped with an image processing unit that generates
The image processing unit cuts out the same region of interest on the furnace wall surface from at least two images taken at different timings from among the series of furnace interior images and the generated furnace wall surface images, and based on the cut out partial images. generate an inspection image of the region of interest ,
The image processing unit includes a machine learning discriminator that estimates whether the input furnace interior image or a partial image thereof is effective for damage inspection of the furnace wall surface,
The image processing unit sequentially inputs the cut out partial images to the discriminator, selects a partial image effective for damage inspection of the furnace wall surface based on the output from the discriminator, and selects a partial image effective for damage inspection of the furnace wall surface. An image processing device that generates an inspection image of the region of interest from a partial image . Receive a series of furnace interior images taken sequentially by the camera as perspective images of the furnace wall surface while moving the camera within the coke oven, and based on the series of furnace interior images, a furnace wall surface image including at least a portion of the furnace wall surface. Equipped with an image processing unit that generates
The image processing unit cuts out the same region of interest on the furnace wall surface from at least two images taken at different timings from among the series of furnace interior images and the generated furnace wall surface images, and based on the cut out partial images. generate an inspection image of the region of interest,
The image processing unit includes a machine learning image generator that synthesizes an effective image that is effective for damage inspection of the furnace wall surface from an input group of in-furnace images or a group of partial images thereof,
The image processing device is characterized in that the image processing unit inputs the cut out partial images to the image generator, synthesizes the effective images, and generates an inspection image of the region of interest from the effective images.

Images