JP2016060931A - Blast furnace tuyere condition observation method and blast furnace tuyere condition observation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately observe a condition near a blast furnace tuyere independently from fluctuation in a blast furnace tuyere image due to fire.SOLUTION: The present invention is a method for observing a condition in a blast furnace tuyere by utilizing a thermal radiation intensity image taken by an imaging apparatus arranged on the blast furnace tuyere, comprising: a step of identifying an ellipse shape matching with an outline shape by applying ellipse shapes to an outline shape of a tuyere from each of thermal radiation intensity images; a step of carrying out geometric transformation to each of thermal radiation intensity images to generate a normalized image and carrying out polar conversion to the normalized image with ellipse parameters of the identified ellipse shape so that each of outline shapes to be a normalized circle having a determined center position and a determined radius; a step of generating a binarized image by binarizing the normalized image after polar conversion generated; and a step of generating light part distribution information showing distribution in the radial direction of a normalized circle of light parts present in the generated binarized image by utilizing the binarized image generated.SELECTED DRAWING: Figure 5

Description

  The present invention relates to a blast furnace tuyere state observation method and a blast furnace tuyere state observation apparatus.

  A blast furnace is a complex and large-scale reaction vessel in which a reduction reaction composed of a plurality of chemical reactions proceeds using iron ore or coke as a raw material. A blast furnace is a type of vertical furnace with a cylindrical bottle shape, with raw materials charged from the top of the furnace (top of the furnace) and hot air (high temperature) supplied from the tuyere provided below the furnace. The reducing gas produced by the air) is used as a raw material, and the hot metal (molten pig iron) is continuously discharged from the outlet at the bottom of the blast furnace by the reduction reaction. The raw material charging from the furnace top is intermittently performed, and each raw material layer in the blast furnace gradually moves downward. Thus, in the continuous operation in the blast furnace, the reaction proceeds in the presence of solid, liquid, and gas.

  Here, predicting the reduction reaction proceeding in the blast furnace is important for stable operation. Therefore, it has been performed to visually check the inside of the blast furnace from the tuyere, but it is difficult to continuously and quantitatively monitor many tuyere by visual monitoring. There was a problem.

  In order to solve such a problem, it is possible to continuously analyze the state of the furnace later by recording an image of the inside of the furnace captured from the blast furnace tuyeres with video. However, since the point that a person determines the state in the furnace is the same, there still remains a problem of lack of quantitativeness. Furthermore, it takes a huge amount of time for a person to analyze a long-time moving image while reproducing it. The problem of needing arises.

  Therefore, as shown in Patent Document 1 below, an index representing the blast furnace state is calculated by performing predetermined image processing on the blast furnace tuyere image obtained by imaging the blast furnace tuyere, and the time series change of the index is monitored. A method has been proposed.

  Here, as an abnormal condition that occurs in the blast furnace, the events that you want to distinguish from the blast furnace tuyere images are the phenomenon called “green ore dropping” in which black unmelted raw material falls randomly in the image, and fine powder blown from the PC nozzle. This is an event called “pulverized coal expansion” in which the charcoal expands from the center toward the circumference. For this reason, in the technique disclosed in Patent Document 1 below, a fixed ellipse is applied to a tuyere image, converted into a normalized circle, and then separated into a circumferential portion and a central portion by polar coordinate conversion. Are distinguished.

JP 2013-185234 A JP 2006-84473 A

  The problem in the event discrimination process disclosed in Patent Document 1 is that the size and center position of the blast furnace tuyere changes from moment to moment due to the heat (temperature difference between the inside of the furnace and the hot air blown in). It is to do. Because a fixed ellipse obtained by performing ellipse approximation processing on the blast furnace tuyere image at rest is applied to the tuyere image, this change causes a normalization circle or subsequent polar coordinate transformation. This causes an error, especially when detecting “pulverized coal expansion”.

  In order to correct the fluctuation of the blast furnace tuyere image due to the above-mentioned hot flame, it is important to extract a contour line corresponding to the blast furnace tuyere from the blast furnace tuyere image before the correction. As a technique for extracting a contour line from certain image data, for example, a technique disclosed in Patent Document 2 has been proposed. In this Patent Document 2, a plurality of collation areas are placed in a section with respect to the contour portion of a model to be matched, and the coordinates of the collation areas are converted into parallel, rotated, and similarity, and then the luminance values of the respective collation areas after conversion, etc. The degree of separation obtained from (1) is calculated, a transformation matrix that maximizes the degree of separation obtained is selected by exhaustive search, and model parameters in the image are determined from parameters of the transformation matrix.

  However, in the method disclosed in Patent Document 2, since the collation area is provided in a sectioned manner, there is a problem that the information held by the contour line is aggregated for each collation area, resulting in insufficient accuracy. Further, since the method disclosed in Patent Document 2 performs exhaustive search in the parameter space, if the lattice size of the parameter space is reduced in order to increase parameter determination accuracy, the calculation time is greatly increased. There is.

  Furthermore, in the contour extraction method based on a normal dynamic contour model such as Snake, in order to give generality to the processing, not a contour of a curve represented by a parameter but an indefinite contour represented by a point cloud ( Non-parametric). Therefore, this technique cannot be applied to the purpose of obtaining the ellipse parameter with high accuracy in order to improve the accuracy of the polar coordinate conversion as described above.

  Therefore, the present invention has been made in view of the above problems, and the object of the present invention is to observe the state near the blast furnace tuyere more accurately without depending on the fluctuation of the blast furnace tuyere image due to the heat. An object of the present invention is to provide a blast furnace tuyere state observation method and a blast furnace tuyere state observation apparatus that can be used.

  In order to solve the above problems, according to an aspect of the present invention, there is provided a method for observing a state in a tuyere using a thermal radiance image captured by an imaging device provided in a blast furnace tuyere. The elliptic shape is applied to the contour shape of the tuyere in each of the acquired thermal radiance images, and the elliptic shape specifying step for specifying the elliptic shape that matches the contour shape is specified in the elliptic shape specifying step. Using the ellipse parameters of the ellipse shape, geometric transformation is performed on the respective thermal radiance images so that each contour shape becomes a normalized circle having a constant center position and a constant radius, and An image conversion step of generating a normalized image and performing a polar coordinate conversion on the normalized image, and the normalized image after the polar coordinate conversion generated in the image conversion step. A binarized image generating step for binarizing and generating a binarized image, and using the generated binarized image, the radial direction of the normalized circle of the bright portion existing in the binarized image A blast furnace tuyere state observation method is provided that includes a bright part distribution information generation step of generating bright part distribution information indicating a distribution at.

  The ellipse parameters of the specified ellipse shape are the center position coordinates of the ellipse shape, the long axis direction length of the ellipse shape corresponding to the image horizontal direction of the thermal radiance image, and the image vertical direction of the thermal radiance image. It is preferable that the length is an elliptical short-axis direction corresponding to.

  In the ellipse shape specifying step, an evaluation function including a line integral of the contour strength located on the arc of the elliptical shape to be applied is defined for the contour strength image in which the image luminance value of the contour portion of the tuyere is emphasized, and the evaluation It is preferable to identify an elliptical shape that matches the contour shape by solving an optimization problem that identifies an elliptical shape that gives the maximum value of the function.

  In the ellipse shape specifying step, a value obtained by subtracting the arc length of the ellipse from a constant multiple of the line integral of the contour intensity is used as the evaluation function, and an elliptical shape surrounding the outside of the tuyere is used as the initial value of the elliptical shape. It is preferable to execute an optimization process that sets and identifies an elliptical shape that gives the maximum value by a gradient method.

  In the elliptical shape specifying step, two types of differential images are generated by differentiating the thermal radiance image in the horizontal direction and the vertical direction of the image, respectively, and the two types of differential images are squared and added together. It is preferable that a contour strength image is generated in step S1 and the elliptical shape is specified based on the contour strength image.

  In the elliptical shape specifying step, prior to the generation of the differential image, the thermal radiance image is subjected to a two-dimensional Gaussian filter process, and the thermal radiance image after the two-dimensional Gaussian filter process is used, A differential image is preferably generated.

  In the elliptical shape specifying step, it is preferable that the contour intensity image is generated after a saturation process or a binarization process is performed on the thermal radiance image.

  In the ellipse shape specifying step, the initial value of the optimization process for the first thermal radiance image of each of the thermal radiance images, and the next optimization process when the optimization process fails The initial value is an oval shape defined in advance, the optimization process for the first thermal radiance image, and the optimization process of the previous optimization process other than the next optimization process when the optimization process fails A value obtained by enlarging the elliptical length in the major axis direction and the minor axis direction length by a constant factor may be used as the initial value.

  In the bright part distribution information generation step, pixel values corresponding to a plurality of pixels having the same radial position and different declination positions are provided for each radial position in the binarized image. It is good also as an element of the said bright part distribution information in the said radial direction position which integrates | accumulates and obtained.

  The blast furnace tuyere state observation method uses the generated bright portion distribution information to calculate a feature amount that characterizes the bright portion distribution information at a certain time, and the calculated feature amount The feature amount coordinate system specified by using the feature amount is set using the feature amount coordinate system specified by the combination of the feature amounts corresponding to the bright part distribution information at a plurality of different times. A trajectory information generating step for generating trajectory information indicating a trajectory associated with time transition of the point, and a state determining step for determining a state in the tuyere using the generated trajectory information, In the feature amount calculating step, the number-of-elements feature amount characterizing the number of the elements included in the predetermined radial direction range and the radial position feature amount indicating the radial position giving the maximum value of the elements are reduced. When Calculation may be.

  In order to solve the above-described problem, according to another aspect of the present invention, an imaging device that is provided at a blast furnace tuyere and that captures a thermal radiance image at the blast furnace tuyere is captured by the imaging device. An arithmetic processing device that performs image processing on each of the thermal radiance images, and the arithmetic processing device has an elliptical shape in the contour shape of the tuyere in each of the acquired thermal radiance images. The elliptical shape specifying unit that specifies an elliptical shape that fits the contour shape, and the elliptical parameters of the elliptical shape that are specified by the elliptical shape specifying unit, each contour shape has a constant center position and a constant radius. The geometrical transformation is performed on each of the thermal radiance images so as to form a normalized circle, and a normalized image is generated, and polar coordinate transformation is performed on the normalized image. An image conversion unit, a binarized image generation unit that generates a binarized image by binarizing the normalized image after the polar coordinate conversion generated by the image conversion unit, and the generated binarization A blast furnace tuyere state comprising: a bright part distribution information generating unit for generating bright part distribution information indicating a radial distribution of the normalized circle of the bright part existing in the binarized image using the image An observation device is provided.

  As described above, according to the present invention, an elliptical shape is applied to the contour shape of the tuyere, an elliptical shape that matches the contoured shape is specified, and each contoured shape is determined using the specified elliptical elliptic parameter. Because the geometrical transformation is performed on each thermal radiance image so that the center position is constant and the radius is constant, the state near the blast furnace tuyere is not dependent on the fluctuation of the blast furnace tuyere image due to the heat. Can be observed with higher accuracy.

It is explanatory drawing for demonstrating a blast furnace. It is explanatory drawing which showed the structure of the blast furnace tuyere state observation apparatus which concerns on the 1st Embodiment of this invention. It is explanatory drawing shown about the imaging device with which the blast furnace tuyere state observation apparatus which concerns on the same embodiment is provided. It is explanatory drawing which showed the example of the thermal radiance luminance image which concerns on the same embodiment. It is the block diagram shown about the structure of the image processing part of the arithmetic processing apparatus with which the blast furnace tuyere state observation apparatus which concerns on the same embodiment is provided. It is the block diagram shown about the structure of the ellipse shape specific | specification part which the image processing part of the arithmetic processing apparatus which concerns on the same embodiment has. It is explanatory drawing for demonstrating the ellipse shape specific process which concerns on the embodiment. It is explanatory drawing for demonstrating the ellipse shape specific process which concerns on the embodiment. It is explanatory drawing for demonstrating the ellipse shape specific process which concerns on the embodiment. It is explanatory drawing for demonstrating the ellipse shape specific process which concerns on the embodiment. It is explanatory drawing for demonstrating the ellipse shape specific process which concerns on the embodiment. It is explanatory drawing for demonstrating the image conversion part which concerns on the same embodiment. It is explanatory drawing for demonstrating the binarized image which concerns on the same embodiment. It is explanatory drawing for demonstrating the bright part distribution information which concerns on the embodiment. It is explanatory drawing for demonstrating the bright part distribution information which concerns on the embodiment. It is explanatory drawing for demonstrating the bright part distribution information which concerns on the embodiment. It is explanatory drawing for demonstrating the feature-value calculation part which concerns on the same embodiment. It is explanatory drawing for demonstrating the locus | trajectory information which concerns on the same embodiment. It is explanatory drawing for demonstrating the judgment method of a blast furnace tuyere state concerning the embodiment. It is the flowchart which showed an example of the flow of the blast furnace tuyere state observation method which concerns on the embodiment. It is the flowchart which showed an example of the flow of the ellipse shape identification method which concerns on the embodiment. It is the block diagram which showed an example of the hardware constitutions of the arithmetic processing unit which concerns on the same embodiment. It is explanatory drawing for demonstrating an experiment example. It is a graph for demonstrating an experiment example. It is a graph for demonstrating an experiment example. It is a graph for demonstrating an experiment example.

  Exemplary embodiments of the present invention will be described below in detail with reference to the accompanying drawings. In addition, in this specification and drawing, about the component which has the substantially same function structure, duplication description is abbreviate | omitted by attaching | subjecting the same code | symbol.

(Outline of the reaction proceeding in the blast furnace)
First, with reference to FIG. 1, the blast furnace to which attention is paid in the embodiment of the present invention will be briefly described. FIG. 1 is an explanatory diagram for explaining a blast furnace to which attention is paid in an embodiment of the present invention.

  As shown in FIG. 1, the blast furnace is a kind of vertical furnace having a cylindrical virtue shape. From a raw material supplied from the top (furnace top) of the furnace and a tuyere provided below the furnace. It functions as a reaction device that reacts with the reducing gas generated by the supplied hot air.

  Examples of raw materials supplied from the top of the furnace include iron oxides such as iron ore and sintered ore, coke, and limestone. Iron ore is a source of iron for pig iron produced by reactions in the blast furnace, and coke not only functions as a heat source for melting iron ore reducing agent and raw materials, but also provides air permeability in the blast furnace. Have a role to hold. Limestone functions as a solvent added to react with the gangue component of iron ore to produce slag having a low melting point and good fluidity.

  As shown in FIG. 1, layers of iron ore (and limestone) and layers of coke are alternately stacked inside the blast furnace. These raw materials move downward in the furnace while maintaining the laminated state as shown in FIG.

  In addition, pulverized coal that functions as a supplementary reducing agent for hot air and coke is supplied from the tuyere shown in FIG. In a region called a raceway near the tuyere, pulverized coal and coke are gasified by combustion by the supplied hot air, and a high-temperature reducing gas composed of carbon monoxide, hydrogen, or the like is generated. This high-temperature reducing gas blows up to the top of the furnace as an ascending current that moves in the furnace. The iron ore in the furnace is reduced by this reducing gas (indirect reduction), and further changes from solid to liquid by the heat of the reducing gas. The iron component that has become liquid is further reduced by the carbon of the coke while dropping in the coke layer (direct reduction), and becomes molten iron containing about 5% of carbon.

  The cohesive zone shown in FIG. 1 is a portion where solid coke exists in the form of slits between the iron components in a semi-molten state, and the phase change of iron as described above mainly occurs in this cohesive zone. Has occurred.

  Thus, in a reactor called a blast furnace, the reaction proceeds in the presence of solid, liquid, and gas. In order to perform stable operation, it is important to predict the reduction reaction that is proceeding in the blast furnace. In the embodiment of the present invention described below, as an index for grasping the situation in the blast furnace, whether or not the phenomenon of “Ore fallen” has occurred and whether or not the phenomenon of “pulverized coal expansion” has occurred Attention is focused on at least two indicators of whether or not.

  “Occurrence of raw ore” is a phenomenon in which unmelted ore falls, and the occurrence of such a phenomenon means that the amount of heat in the blast furnace is insufficient. Therefore, when a raw ore drop occurs, it is necessary to take measures to increase the amount of heat in the furnace, such as increasing the amount of coke.

  “Pulverized coal expansion” is a phenomenon in which an image of unburned pulverized coal that normally occupies about 1/3 of an image of the tip of a nozzle in a captured image obtained by imaging a tuyere rapidly expands. The occurrence of such a phenomenon suggests that the shape of the raceway has changed from a preferable shape, and therefore, a measure for making the shape of the raceway in a good state is necessary.

(About blast furnace tuyere state observation equipment)
First, the blast furnace tuyere state observation apparatus 10 according to the embodiment of the present invention will be described in detail with reference to FIG.

  As shown in FIG. 2, the blast furnace tuyere state observation device 10 according to the present embodiment includes an imaging device 100 and an arithmetic processing device 200.

<About the imaging device>
The imaging device 100 is a device that captures the distribution of thermal radiance at the tuyere and generates a thermal radiance image. The imaging apparatus 100 includes various optical elements such as a lens and an imaging element such as a CCD (Charge Coupled Device) or a CMOS (Complementary Metal Oxide Semiconductor). Here, the imaging apparatus 100 according to the present embodiment may be capable of generating a still image or may be capable of generating a moving image. Further, the imaging apparatus 100 according to the present embodiment may be capable of capturing a monochrome image or may be capable of capturing a color image. Note that when an imaging device capable of capturing a color image is used, a one-channel luminance image may be generated. That is, as the luminance image generating means, only one of R, G, and B among the RGB components may be used, or conversion from the RGB color space to the YCbCr color space is performed, and only the Y component is used. May be used.

  The imaging device 100 is controlled by an arithmetic processing device 200 described later, and a trigger signal for imaging is output from the arithmetic processing device 200 at every predetermined frame rate. The imaging device 100 captures thermal radiation from the tuyere in response to the trigger signal output from the arithmetic processing device 200 and outputs the generated thermal radiance image to the arithmetic processing device 200.

  FIG. 3 is an explanatory diagram for explaining an installation state of the imaging apparatus 100 according to the present embodiment, and FIG. 4 is an explanatory diagram illustrating an example of a thermal radiance image according to the present embodiment. Although the blast furnace is covered with heat-resistant bricks, as shown in FIG. 3, pulverized coal is supplied near the tuyere by a PC lance and hot air of about 1200 ° C. is supplied. A raceway is formed in a range of about 1 m from the tuyere, and in this raceway, the temperature is higher than 2000 ° C. due to combustion of pulverized coal, coke and the like.

  The imaging apparatus 100 according to the present embodiment is installed in an observation window for observing the state of the tuyere, and images thermal radiation from the raceway near the tuyere to obtain a thermal radiance image. The tuyere is usually circular, but the imaging device 100 takes an image so that the tuyere is looked down obliquely from above, so the tuyere shape of the captured thermal radiance image is as shown in the upper left of FIG. Moreover, it becomes a substantially elliptical shape.

  When it is determined that the operating state of the blast furnace is good, the imaging device 100 has an imaging field of view, a focus, and the like adjusted in advance, and appropriate imaging processing is performed. When the blast furnace tuyere is in good condition, as shown in the upper right of FIG. 4, the tip of the PC lance is reflected in the field of view, and the pulverized coal supplied from the tip of the PC lance is It will occupy about 1/3. Further, in areas other than the PC lance and pulverized coal, heat radiation caused by the raceway temperature is reflected.

  When a raw ore drop occurs, as shown schematically in the lower left of FIG. 4, the ore that falls is reflected, so that the entire field of view temporarily becomes dark and the thermal radiance decreases. Further, when pulverized coal expansion occurs, as schematically shown in the lower right of FIG. 4, the area occupied by the pulverized coal rapidly expands from the lower right to the upper left of the field of view, and then in the upper right of FIG. A characteristic pattern of recovery to the state shown is observed.

  The example of the thermal radiance image according to the present embodiment has been specifically described above with reference to FIG.

<Overall configuration of arithmetic processing unit>
Next, returning to FIG. 2 again, the overall configuration of the arithmetic processing apparatus 200 according to the present embodiment will be described.

  The arithmetic processing device 200 according to the present embodiment performs image processing on the thermal radiance image captured by the imaging device 100 to generate bright part distribution information described later. Moreover, the arithmetic processing unit 200 can also determine the state of the tuyere (that is, the occurrence of raw mining or pulverized coal expansion) based on the generated bright part distribution information.

  As illustrated in FIG. 2, the arithmetic processing device 200 mainly includes an imaging control unit 201, an image processing unit 203, a display control unit 205, and a storage unit 207.

  The imaging control unit 201 is realized by a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a communication device, and the like. The imaging control unit 201 performs tuyere imaging control by the imaging apparatus 100 according to the present embodiment. More specifically, the imaging control unit 201 sends a control signal for starting imaging to the imaging apparatus 100 when imaging of the thermal radiance image of the tuyere is started.

  The image processing unit 203 is realized by, for example, a CPU, a ROM, a RAM, a communication device, and the like. The image processing unit 203 performs image processing as described below on the imaging data of the thermal radiance image acquired from the imaging device 100, and generates bright part distribution information described later. Further, the image processing unit 203 determines whether or not a raw mineral drop or pulverized coal expansion has occurred based on the generated bright part distribution information. The image processing unit 203 transmits the generated bright part distribution information and information related to the determination result such as raw mining loss and pulverized coal expansion to the display control unit 205.

  The image processing unit 203 will be described in detail later again.

  The display control unit 205 is realized by, for example, a CPU, a ROM, a RAM, an output device, and the like. The display control unit 205 outputs the bright part distribution information and the determination result of the raw mine dropping / pulverized coal expansion transmitted from the image processing unit 203 to an output device such as a display provided in the arithmetic processing device 200 or the arithmetic processing device 200. Display control when displaying on an external output device or the like is performed. Thereby, the user of the blast furnace tuyere state observation apparatus 10 can grasp the bright part distribution information and information on the state of the blast furnace tuyere on the spot.

  The storage unit 207 is realized by, for example, a RAM or a storage device included in the arithmetic processing device 200 according to the present embodiment. The storage unit 207 stores various parameters, intermediate progress of processing, and various databases and programs that need to be saved when the arithmetic processing apparatus 200 according to the present embodiment performs some processing, as appropriate. To be recorded. The storage unit 207 can be freely read and written by the imaging control unit 201, the image processing unit 203, the display control unit 205, and the like.

[About image processing unit]
Next, the image processing unit 203 included in the arithmetic processing apparatus 200 according to the present embodiment will be described in detail with reference to FIGS.
FIG. 5 is a block diagram illustrating the configuration of the image processing unit of the arithmetic processing device included in the blast furnace tuyere state observation device according to the present embodiment, and FIG. 6 is the image processing unit of the arithmetic processing device according to the present embodiment. It is the block diagram shown about the structure of the ellipse shape specific | specification part which has. FIG. 7A to FIG. 10 are explanatory diagrams for explaining the elliptical shape specifying process according to the present embodiment. FIG. 11 is an explanatory diagram for explaining an image conversion unit according to the present embodiment, and FIG. 12 is an explanatory diagram for explaining a binarized image according to the present embodiment. 13 to 15 are explanatory diagrams for explaining the bright portion distribution information according to the present embodiment. FIG. 16 is an explanatory diagram for describing the feature amount calculation unit according to the present embodiment, FIG. 17 is an explanatory diagram for describing the trajectory information according to the present embodiment, and FIG. 18 is the present embodiment. It is explanatory drawing for demonstrating the judgment method of the blast furnace tuyere state which concerns on a form.

  As illustrated in FIG. 5, the arithmetic processing device 200 according to the present embodiment mainly includes an elliptical shape specifying unit 211, an image conversion unit 213, and an index calculation unit 215. Further, as shown in FIG. 5, the index calculation unit 215 includes a binarized image generation unit 217, a bright part distribution information generation unit 219, a feature amount calculation unit 221, a trajectory information generation unit 223, and a state determination. Part 225.

○ Oval shape specifying part 211
The elliptical shape specifying unit 211 is realized by, for example, a CPU, a ROM, a RAM, and the like. The elliptical shape specifying unit 211 applies an elliptical shape to the contour shape of the tuyere in each thermal radiance image acquired by the imaging device 100, and specifies an elliptical shape that matches the contour shape.

  As shown in FIG. 6, the elliptical shape specifying unit 211 includes a saturation processing unit 231, a contour extraction unit 233, and a shape calculation unit 235.

  The saturation processing unit 231 is realized by, for example, a CPU, a ROM, a RAM, and the like. The saturation processing unit 231 performs a saturation process as illustrated in FIG. 7A or a binarization process as illustrated in FIG. 7B on the thermal radiance image generated by the imaging apparatus 100, and the thermal radiance image The gradient of the luminance value is made constant.

  When the generated thermal radiance image is used as it is for subsequent image processing, if a “green ore drop” has occurred in the blast furnace, the gradient of the luminance value in the thermal radiance image is reduced and the contour is It is often unclear. As will be described below, in the elliptical shape specifying process by the elliptical shape specifying unit 211, a differential process is performed on the thermal radiance image. However, if the gradient of the luminance value is small, the differential value is small and the process is correct. May not be done. Therefore, the saturation processing unit 231 performs a saturation process or a binarization process as described below on the thermal radiance image so that the output value by the subsequent differentiation process is stabilized.

  FIG. 7A is a graph schematically showing the relationship between the input luminance value and the output luminance value when the saturation processing unit 231 performs the saturation processing. In the saturation processing as shown in FIG. 7A, the saturation processing unit 231 refers to the pixel value (that is, the input luminance value) of the thermal radiance image to be processed, and the luminance value of the pixel of interest is the luminance. When it is less than the value Th, the output luminance value proportional to the input luminance value is associated as the luminance value of the pixel of interest. When the luminance value of the pixel of interest is equal to or higher than the luminance value Th, the saturation processing unit 231 sets the output luminance value to a certain value (hereinafter referred to as “input luminance value”). , Also referred to as saturation value). This saturation value is set to a value larger than the luminance value of the portion that is not the tuyere (for example, the end of the imaging field in the thermal radiance image schematically shown in FIG. 4) in the thermal radiance image. And make the outline of the tuyere clear. As a result, in the thermal radiance image, the region in which the space in the blast furnace is imaged is represented, for example, as white, and the outline of the tuyere portion is clarified.

  FIG. 7B is a graph schematically illustrating the relationship between the input luminance value and the output luminance value when the saturation processing unit 231 performs binarization processing. In the case of the binarization process as illustrated in FIG. 7B, the saturation processing unit 231 refers to the pixel value of the thermal radiance image to be processed, and the luminance value of the pixel of interest is less than the luminance value Th. In this case, the output luminance value is set to zero. When the luminance value of the pixel of interest is equal to or higher than the luminance value Th, the saturation processing unit 231 sets the output luminance value to a certain saturated value regardless of the magnitude of the input luminance value. To be.

  When the saturation processing unit 231 performs the saturation processing or the binarization processing on the thermal radiance image, the saturation processing unit 231 outputs the obtained data of the processed thermal radiance image to the contour extraction unit 233 described later. To do.

  The contour extraction unit 233 is realized by, for example, a CPU, a ROM, a RAM, and the like. The contour extraction unit 233 uses the thermal radiance image output from the saturation processing unit 231 to extract the tuyere contour from the thermal radiance image, and emphasizes the strength (image luminance value) of the contour part. Generated contour intensity image.

  More specifically, the contour extraction unit 233 generates two types of differential images by differentiating the thermal radiance image in the image horizontal direction and the image vertical direction, respectively, and squares the obtained two types of differential images, respectively. Then, by adding together, a contour strength image in which the strength of the contour portion of the tuyere is increased is generated. Here, prior to the generation of the differential image, the contour extraction unit 233 preferably performs a two-dimensional Gaussian filter process on the thermal radiance image to blur the contour. As a result, the contour line can be thickened, and the elliptical shape search process corresponding to the contour in the shape calculation unit 235 described later can be made easier.

  Hereinafter, the contour extraction processing in the contour extraction unit 233 will be specifically described with reference to FIG.

Now, it is assumed that the thermal radiance image as shown at the left end of FIG. 8 is output from the saturation processing unit 231. Here, the image vertical direction of the thermal radiance image is referred to as the X-axis direction for convenience, and the image horizontal direction is referred to as the Y-axis direction for convenience. The contour extraction unit 233 first applies a Gaussian filter (filter using a two-dimensional Gaussian function represented by exp {− (x ² + y ² ) / σ ² } as a kernel function) to the thermal radiance image. The contour blurring process is performed by acting. As a result, as shown in FIG. 8, the thermal radiance image can be blurred over the whole, and the contour line corresponding to the tuyere-shaped outer shape can be thickened.

  Here, the parameter stipulating the spread of the two-dimensional Gaussian function (in other words, the parameter indicating the thickness of the contour line after thickening) σ is thicker as the optimization calculation in the shape calculation unit 235 described later functions. The value is set such that the contour line thus formed does not protrude from the image. Such a value is not particularly limited, and may be appropriately set according to the accuracy required for the optimization calculation described later, the image size of the thermal radiance image, and the like. For example, the value is set to about σ = 10 pixels. be able to.

  Next, the contour extraction unit 233 performs differential processing in the X direction on the thermal radiance image after filtering to generate a differential image in the X direction, and for the thermal radiance image after filtering. Y-direction differential processing is performed to generate a Y-direction differential image. As a result, as shown in FIG. 8, two types of differential images are generated from one thermal radiance image. In the differential image shown in FIG. 8, the portion where the luminance value = 0 is displayed in gray, the portion where the luminance value> 0 is displayed in white, and the portion where the luminance value <0 is displayed in black. ing. With this differential image, an edge portion in the X direction and an edge portion in the Y direction in the thermal radiance image after the filter processing are detected.

  Subsequently, the contour extracting unit 233 adds each of the two types of differential images to the square and adds them to obtain a contour strength image. As a result, as shown at the right end of FIG. 8, a contour strength image is generated in which the strength of the pixel corresponding to the tuyere-shaped contour line is enhanced. Note that FIG. 8 shows a normalized pixel in which the luminance value of the pixel corresponding to the contour line is 1 and the luminance value of the pixels other than the contour line is 0.

  The contour extraction unit 233 outputs the contour strength image generated in this way to the shape calculation unit 235.

  The shape calculation unit 235 is realized by, for example, a CPU, a ROM, a RAM, and the like. The shape calculation unit 235 uses the contour strength image generated by the contour extraction unit 233 to calculate an ellipse parameter representing an elliptical shape fitted to the tuyere contour shape.

  More specifically, the shape calculation unit 235 searches for an elliptical elliptical parameter (ellipse center position, elliptical long axis, and elliptical short axis size) that best fits the contour line represented by the contour strength image. In such a search, as schematically shown in the left diagram of FIG. 9, an ellipse that surrounds the outside of the tuyere shape is set as an initial ellipse that is an initial value, and then the ellipse is gradually contracted. After that, when the ellipse touches the contour line represented by the contour strength image, the search process ends.

  In order to realize such a search process, the shape calculation unit 235 includes an evaluation function represented by the following expression 101 including a line integral of the contour strength positioned on the elliptical arc to be applied to the target contour strength image. E is specified. Then, the shape calculation unit 235 solves the optimization problem “search for an elliptical shape that gives the maximum value of the evaluation function E” by the gradient method, and obtains the elliptical shape corresponding to the obtained solution as the contour shape of the tuyere An elliptical shape that conforms to

Here, in Equation 101 above, the integral symbol is a line integral on the arc of the ellipse determined by the ellipse parameters,
E: Evaluation function for obtaining a maximum value I _e : Sum of contour strengths (luminance values) located on an elliptical arc to be fitted K: Constant for determining the balance between the force to reduce the elliptical shape and the attractive force to the contour line .

  The first term on the right side of the above equation 101 is obtained by multiplying the sum of the brightness of the contour intensity image along the arc of the ellipse by a constant, and the value of the first term on the right side increases, which means that the ellipse arc matches the contour. Indicates that The second term on the right-hand side is one in which the sign of the arc length of the ellipse is minus because -1 is linearly integrated, and the value of the second term increases, that is, the absolute value of the second term decreases. Corresponds to the contraction of the ellipse. Since the constant K depends on the arc length of the ellipse and the brightness value of the contour intensity image, it is preferable to experimentally determine the constant K by prior verification using past operation data or the like. The specific value of the constant K is not particularly limited, but can be set to K = 20.

  Here, for the purpose of fitting an elliptical shape to the contour line, it may seem that only the first term on the right side of the equation 101 is sufficient. However, when a large ellipse as shown in the left diagram of FIG. 9 is set as the initial value of the optimization problem, the value of the line integral defined only by the first term on the right side of the above equation 101 is used as the objective function. As schematically shown in the right diagram of FIG. 9, the gradient of the objective function in the vicinity of the initial value becomes small or sometimes becomes zero. As a result, when only the first term on the right side is the objective function, the search for the maximum value cannot be started by the gradient method. Therefore, in the present embodiment, the second term on the right side of the equation 101 is added for the purpose of giving the gradient method driving force in the direction in which the ellipse contracts. As a result, as schematically shown in FIG. 10, the gradient of the objective function (that is, the evaluation function E) becomes large even in the vicinity of the initial value, and the local maximum search by the gradient method can be performed.

  Further, when solving the optimization problem, it is possible to start from a sufficiently small ellipse and gradually increase the size of the ellipse (the direction in which the search starts from the left end of the graph in the right diagram of FIG. 9). However, as shown in FIG. 4 and FIG. 8 left figure, the tuyere contour shape according to this embodiment is divided into two regions. Another local solution as shown may be obtained. Therefore, in this embodiment, a policy is adopted in which a search is started from an ellipse parameter that is surely outside the outline shape of the tuyere from past performance data, and the first local solution is obtained while the ellipse is contracted. .

  The details of the gradient method used as a solution for the optimization problem are not particularly limited. For example, the non-patent document “Optimization Method” (Toshihide Ibaraki, Masao Fukushima, Kyoritsu Shuppan, page 109) Known ones can be used as described.

The shape calculation unit 235 identifies the elliptical shape that gives the maximum value of the evaluation function E by the above-described method, the identified elliptical center position (X _C , Y _C ), and the elliptical long axis size 2a. And the size 2b of the elliptical minor axis are calculated. When the shape calculation unit 235 calculates the above ellipse parameters (the size of the center position, the ellipse straight axis, and the ellipse minor axis) for each contour strength image, the obtained ellipse parameter is converted into an image conversion unit described later. To 213.

  Here, the shape calculation unit 235 stores the ellipse surrounding the outer side of the tuyere shape used in the above method (initial ellipse) as a fixed value determined by prior verification using past operation data in the storage unit 207 or the like. Store it. In addition, the shape calculation unit 235 includes (A) an optimization initial value when the optimization problem is first solved for the thermal radiance image acquired as the first frame, and (B) an optimization in the optimization problem. The initial value of the next optimization process when the optimization process fails without convergence and the elliptical shape corresponding to the “solution to be obtained” shown in FIG. A predetermined elliptical shape).

  In addition, during the optimization process other than the above (A) and (B), the shape calculation unit 235 expands the major axis direction length and the minor axis direction length of the elliptical shape, which is the solution of the previous optimization process, by a constant multiple. The obtained value (for example, 1.05 to 1.10 times) is set as the initial value. As a result, the number of searches for an elliptical shape that gives a maximum value can be reduced, and the calculation load when calculating the elliptical shape can be reduced.

  The configuration of the elliptical shape specifying unit 211 included in the image processing unit 203 according to the present embodiment has been described in detail above with reference to FIGS.

○ Image conversion unit 213
Returning to FIG. 5 again, the image conversion unit 213 included in the image processing unit 203 according to the present embodiment will be described.
The image conversion unit 213 is realized by a CPU, a ROM, a RAM, and the like, for example. The image conversion unit 213 uses the elliptic parameter of the ellipse shape specified by the ellipse shape specification unit 211 so that each contour shape becomes a normalized circle having a constant center position and a constant radius. A geometric transformation is performed on the image to generate a normalized image. Thereafter, the image conversion unit 213 performs polar coordinate conversion on the generated normalized image. Hereinafter, the image conversion process performed by the image conversion unit 213 will be described in detail.

  As schematically illustrated in FIG. 11, the image conversion unit 213 first performs a geometric conversion process on the thermal radiance image using the ellipse parameters specified by the ellipse shape specifying unit 211. Thereby, the elliptical shape which is the outline of a tuyere is normalized to a perfect circle.

More specifically, the image conversion unit 213 determines the center position (X _{C ′} , Y _{C ′} ) of the normalized circle and the radius of the normalized circle that are stored in advance in the storage unit 207 as calculation parameters of the image processing unit 203. Get R. Subsequently, the image conversion unit 213 makes the center position (X _C , Y _C ) of the elliptic shape specified by the elliptic shape specification unit 211 the center position (X _{C ′} , Y _{C ′} ) of the normalized circle. The thermal radiance image of interest is translated.

  Subsequently, as illustrated in FIG. 11, the image conversion unit 213 reduces or enlarges the X axis corresponding to the long axis to (R / a) times and expands the Y axis corresponding to the short axis to (R / b). ) To a normalized circle of radius R. Thereby, the image converter 213 can generate a normalized image from the thermal radiance image.

  Here, as the geometric transformation when the elliptical shape is normalized to a circle, a known one can be used, and for example, affine transformation may be used.

  Subsequently, as illustrated in FIG. 11, the image conversion unit 213 performs polar coordinate conversion on the generated normalized image. By calculating the center position of the normalized circle, the position of each pixel constituting the normalized image can be represented by polar coordinates (r, θ). The image conversion unit 213 performs polar coordinate conversion using the calculated center position as a reference, with the range of the radius r and the range of the deflection angle θ set to 0 ≦ r ≦ R and 0 ° ≦ θ <360 °, respectively. By performing polar coordinate conversion, the normalized image becomes a band-like image as shown in FIG.

  Here, in the belt-like image as shown in FIG. 11, the length of the side parallel to the radial direction r corresponds to the radius R of the normalized circle and is parallel to the declination direction θ. The length of the side corresponds to the circumference of the normalized circle. Of the sides parallel to the declination direction θ, one corresponds to the center of the normalized circle, and the other corresponds to the outer periphery of the normalized circle.

  The image conversion unit 213 outputs data corresponding to the band-like image generated by polar coordinate conversion to the binarized image generation unit 217 of the index calculation unit 215.

○ Indicator calculation unit 215
Returning to FIG. 5 again, the index calculation unit 215 included in the image processing unit 203 according to the present embodiment will be described.
The index calculation unit 215 according to the present embodiment is realized by a CPU, a ROM, a RAM, and the like, for example. The index calculation unit 215 calculates an index representing the state in the tuyere by performing predetermined image processing on the normalized image after the polar coordinate conversion generated by the image conversion unit 213. Hereinafter, the functions of the processing units included in the index calculation unit 215 will be described in detail with reference to FIGS.

○ Binary image generation unit 217
The binarized image generation unit 217 is realized by, for example, a CPU, a ROM, a RAM, and the like. The binarized image generation unit 217 binarizes the belt-like image that is a normalized image after polar coordinate conversion, and generates a binarized image. More specifically, the binarized image generation unit 217 binarizes the band-shaped image by comparing the pixel value (luminance value) of each pixel constituting the band-shaped image with the binarization threshold value, and binarizes the band-shaped image. Generate a digitized image.

  Here, the binarization threshold used by the binarized image generation unit 217 for binarization processing is not a fixed threshold, but the maximum luminance included in the thermal radiance image (and thus the binarized image). The threshold value is changed accordingly. If the thermal radiance image has a luminance value of a certain level or more and the luminance distribution is uniform, the blast furnace tuyere regardless of the temperature (raceway temperature) near the tuyere It is because it can be judged that the state of is good. Specifically, the binarized image generation unit 217 includes (a) a preset luminance threshold as a binarization threshold, and (b) a product obtained by multiplying the highest luminance value by a preset coefficient. , Whichever is larger is used as the binarization threshold. The brightness threshold value and coefficient may be set as appropriate according to the characteristics and operating conditions unique to the blast furnace.For example, a low brightness value corresponding to a low temperature that is clearly not possible as a combustion part in normal operation is used as the brightness threshold value. Set. By setting the binarization threshold in this way, a binary image normalized with respect to fluctuations in the maximum brightness, that is, raceway temperature, is obtained during normal operation, and abnormal conditions such as visual field occlusion due to missing raw minerals, Detection can be performed under the condition that all binary images are zero.

  The binarized image generation unit 217 binarizes the belt-like image using such a binarization threshold, thereby generating a binarized image as shown in FIG. 12, for example. In a binarized image, a pixel having a luminance value equal to or higher than the binarization threshold becomes a portion having a pixel value of 1 (hereinafter also referred to as a bright portion) and has a luminance value less than the binarization threshold. The pixel that has been used is a portion having a pixel value of 0 (hereinafter also referred to as a dark portion).

  The binarized image generation unit 217 outputs data corresponding to the generated binarized image to the bright part distribution information generation unit 219 described later.

○ Bright part distribution information generation unit 219
The bright part distribution information generation unit 219 is realized by, for example, a CPU, a ROM, a RAM, and the like. The bright part distribution information generation unit 219 uses the generated binarized image to generate bright part distribution information indicating the distribution in the radial direction of the normalization circle of the bright part present in the binarized image. .

  More specifically, the bright part distribution information generation unit 219 corresponds to a plurality of pixels having the same radial position and different declination positions for each radial position r in the binarized image. The pixel values to be integrated are integrated, and the obtained integrated value is used as an element of the bright portion distribution information at the corresponding radial position. Such pixel value integration processing corresponds to projecting the binarized image in the direction of the dashed arrow shown in FIG. Further, since the image used by the bright part distribution information generation unit 219 is a binarized image, the pixel value integration result represents the number of pixels corresponding to the bright part at the radial position of interest. Will be. By executing such processing for each position (0 ≦ r ≦ R) in the radial direction r, the bright part distribution information generation unit 219 generates the bright part distribution information as illustrated in FIG. can do. The bright portion distribution information is not limited to the projection value (Σθ), and W (r) Σθ multiplied by the weight W (r) that is a function of the radial direction r may be used as the bright portion distribution information. As the weight W (r), for example, if W (r) = r, the distribution information is multiplied by a weight proportional to the radial position.

  As shown in FIG. 4, a bright portion remains only at the outer peripheral portion of the image when the pulverized coal expands, but there is a feature that the outer peripheral portion and the inner peripheral portion of the image become dark when the raw mineral is dropped. There is a difference in the projection value (Σθ) at the position r.

  When each element constituting the bright part distribution information is plotted for each radial position, a graph as shown in FIG. 13 can be generated. This is a graph showing how many pixels corresponding to a portion exist. Therefore, when the polar coordinate conversion is performed, for example, in increments of 1 ° with respect to the declination direction θ, the generated bright portion distribution information includes bright portions having a luminance value equal to or higher than the binarization threshold at the position of the radius r. This is information indicating how many minutes have existed.

  The bright part distribution information generation unit 219 performs such bright part distribution information generation processing for each captured thermal radiance image. Further, the bright part distribution information generation unit 219 can generate time series transition information indicating the time series transition of the bright part distribution information by arranging the generated bright part distribution information at each time t in order of time. . Specifically, when acquiring the binarized image corresponding to the thermal radiance image at a certain time t, the bright part distribution information generating unit 219 generates the bright part distribution information based on the acquired binarized image, By sequentially storing the generated bright part distribution information in a memory area provided in the storage unit 207 or the like, it is possible to generate time series transition information as described above.

  The arithmetic processing apparatus 200 according to the present embodiment may represent the time-series transition information generated in this way as a three-dimensional graph as shown in FIG. 14, or a projection as shown in FIG. It may be expressed as a two-dimensional graph (shading diagram) in which the shading of the color changes according to the magnitude of the value.

  Here, in the shading diagram shown in FIG. 15, the larger the projection value is, the more white it is displayed. In addition, FIG. 15 also shows the density diagrams in each of the states of the blast furnace tuyere: (a) good, (b) slightly good, (c) raw mineral loss, and (d) pulverized coal expansion. Yes.

  In the good state shown in FIG. 15A, the proportion of the bright portion increases as the outer periphery of the tuyere is approached, and even when time elapses, the proportion of the bright portion in the belt-like image is one. It is like. Further, in the slightly good state shown in FIG. 15B, although the proportion of the bright portion is small, the proportion of the bright portion in the belt-like image is uniform even when time elapses.

  FIG. 15C shows time-series transition information when a raw ore drop occurs in 60 seconds to 80 seconds. As is clear from FIG. 15 (c), almost the entire radial direction is dark at the time when the raw ore drop occurs. On the other hand, FIG. 15D shows time-series transition information when pulverized coal expansion occurs in 10 seconds to 30 seconds. As is clear from FIG. 15 (d), when pulverized coal expansion occurs, the dark part increases toward the periphery of the tuyere, and then the bright part increases toward the center of the tuyere, that is, toward a position in the radial direction that is small. It shows a specific behavior like

  When the bright part distribution information generation unit 219 generates the bright part distribution information and the time-series transition information by the method described above, the bright part distribution information generation unit 219 outputs the generated information to the feature amount calculation unit 221 described later. The bright part distribution information generation unit 219 may output the generated information to the display control unit 205 and display the information on the display screen.

  In the above description, the case where the processing as described above is performed for each generated thermal radiance image has been described. However, it is generated based on the thermal radiance image or the thermal radiance image. At least one of the information may be used after being smoothed with a predetermined time constant. That is, a thermal radiance image, a binarized image, a belt-like image, etc. may be used after being smoothed by exponential smoothing or moving average having a preset time constant, or using an unsmoothed image. The generated bright part distribution information and time series transition information may be smoothed by exponential smoothing or moving average having a preset time constant. The smoothing time constant may be set to about 1/10 of the duration of the phenomenon to be observed. For example, since the expansion of pulverized coal is a phenomenon that continues for about 10 seconds, the time constant is set to about 1 second.

○ Feature amount calculation unit 221
The feature amount calculation unit 221 is realized by, for example, a CPU, a ROM, a RAM, and the like. The feature quantity calculation unit 221 uses the bright part distribution information generated by the bright part distribution information generation unit 219 to calculate a feature quantity that characterizes the bright part distribution information at a certain time.

  More specifically, the feature amount calculation unit 221 includes, for the light portion distribution information at each time, an element number feature amount that characterizes the number of elements included in a predetermined radial direction range, and a radial position that gives the maximum value of the element. And at least a radial position feature quantity indicating Specifically, as illustrated in FIG. 16, the feature amount calculation unit 221 calculates the bright portion average pixel number m within the specified range as the element number feature amount and also outputs the projection value as the radial position feature amount. The highest pixel radius n, which is the radius at which is the maximum, is calculated. Here, the designated range in FIG. 16 may be selected from the range where the shading changes in the shading diagram of FIG.

  Here, the bright part average pixel number m is a feature amount indicating how much bright part is included in the bright part distribution information (in other words, how much dark part is included), and the maximum pixel radius n is This is a feature amount indicating in which part of the binarized image the bright portion remains in the radial direction.

  In the following description, the case where the bright portion average pixel number m within the specified range is calculated as the element number feature amount will be described as an example. However, the maximum pixel within the specified range is used instead of the bright portion average pixel number m. The intermediate value of the number, the number of pixels, the mode value, etc. may be calculated, or the maximum number of pixels, the intermediate value of the number of pixels, the mode value, etc. may be calculated in addition to the bright portion average pixel number m.

  When the feature amount calculation unit 221 calculates the light portion average pixel number m and the maximum pixel radius n for the light portion distribution information at each time, the feature amount calculation unit 221 outputs the calculated feature amounts to the trajectory information generation unit 223.

○ Trajectory information generation unit 223
The trajectory information generation unit 223 is realized by a CPU, a ROM, a RAM, and the like, for example. The trajectory information generation unit 223 uses the feature amount calculated by the feature amount calculation unit 221 to set a feature amount coordinate system defined using the feature amount. After that, the trajectory information generation unit 223 generates trajectory information indicating the trajectory of the point in the feature amount coordinate system specified by the combination of feature amounts corresponding to the bright portion distribution information at a plurality of different times with time transition. To do.

  FIG. 17 shows an example of trajectory information generated by the trajectory information generation unit 223. The combination (m, n) of feature amounts calculated by the feature amount calculation unit 221 indicates where the bright portion exists in the binarized image, and indicates the state near the tuyere at a certain time. It can be said that it is a representative point to represent. Therefore, by representing the time transition of the point (m, n) as a trajectory, it is possible to easily grasp the state change near the tuyere.

  Here, when the state of the blast furnace tuyere is good, the locus of the point represented by (m, n) is concentrated on the upper right of the mn plane, and the state of the blast furnace tuyere is slightly good. In this case, the locus of the point represented by (m, n) is concentrated from the central portion of the mn plane to the slightly lower left region. On the other hand, when a raw ore drop occurs, the locus of the point represented by (m, n) changes from the substantially central part of the mn plane to the vicinity of the origin, and pulverized coal expansion occurs. In this case, the locus of the point represented by (m, n) moves in the horizontal direction in the mn plane.

  As described above, when the trajectory information is generated based on the calculated feature amount, it is understood that a trajectory characteristic of these phenomena is drawn when a raw mining drop or pulverized coal expansion occurs.

  When the trajectory information generation unit 223 generates trajectory information as described above, the trajectory information generation unit 223 outputs the generated trajectory information to a state determination unit 225 described later. Further, the trajectory information generation unit 223 may output the generated trajectory information to the display control unit 205 and display it on the display screen.

○ State determination unit 225
The state determination unit 225 is realized by a CPU, a ROM, a RAM, and the like, for example. The state determination unit 225 determines the state of the blast furnace tuyere based on the trajectory information generated by the trajectory information generation unit 223. More specifically, the state determination unit 225 determines whether or not a raw mine drop or pulverized coal expansion has occurred in the blast furnace tuyere by paying attention to the region where the locus exists.

  As described earlier, there is a great difference in the transition of trajectory information between when the state of the blast furnace tuyere is good and when raw mining or pulverized coal expansion occurs. Therefore, as illustrated in FIG. 18, the state determination unit 225 divides the feature amount coordinate system (mn plane) into a plurality of regions, and assigns labels indicating the state of tuyere in advance to each region. In addition, the state of the tuyere is determined based on in which region the point representing the state exists.

  In the example illustrated in FIG. 18, the mn plane is divided into four regions, and a point (m, n) is present in the upper right region at time t. In this case, the state determination unit 225 determines the state of the tuyere that the combustion state is good because the point (m, n) at time t is in the region labeled “good”. It will be.

  The region classification as shown in FIG. 18 may be performed by classifying the locus of the point (m, n) in the past operation state, and a method for classifying the locus of the point (m, n) is publicly known. Any method can be used.

  When the state determination unit 225 determines the state of the blast furnace tuyere using the method described above, the state determination unit 225 outputs information indicating the determination result to the display control unit 205. Thereby, the user of the blast furnace tuyere state observation apparatus 10 can grasp the determination result regarding the state of the blast furnace tuyere on the spot.

  18 illustrates the case where the mn plane is divided into four regions. However, the number of regions to be divided is not limited to the example illustrated in FIG. 14 and is less than four. There may be four or more.

  Moreover, in the above description, although the case where the state of a blast furnace tuyere was discriminate | determined by dividing a mn plane beforehand and giving a label was demonstrated, the method of judging the state of a blast furnace tuyere is in the said example. It is not limited. For example, a discriminator such as a neural network or a support vector machine (SVM) is generated by learning processing using past feature values m and n and a sensory test result by a tester based on the image data as teacher data. May be used for state determination in the feature quantity mn space.

  The configuration of the image processing unit 203 included in the arithmetic processing device 200 according to the present embodiment has been described above in detail.

  Heretofore, an example of the function of the arithmetic processing apparatus 200 according to the present embodiment has been shown. Each component described above may be configured using a general-purpose member or circuit, or may be configured by hardware specialized for the function of each component. In addition, the CPU or the like may perform all functions of each component. Therefore, it is possible to appropriately change the configuration to be used according to the technical level at the time of carrying out the present embodiment.

  A computer program for realizing each function of the arithmetic processing apparatus according to the present embodiment as described above can be produced and installed in a personal computer or the like. In addition, a computer-readable recording medium storing such a computer program can be provided. The recording medium is, for example, a magnetic disk, an optical disk, a magneto-optical disk, a flash memory, or the like. Further, the above computer program may be distributed via a network, for example, without using a recording medium.

  As described above, in the blast furnace tuyere state observation apparatus 10 according to the present embodiment, the elliptical shape specifying unit 211 performs high-speed and high-precision elliptical fitting processing, and the image conversion unit 213 performs high-precision elliptical-shaped fitting processing. Since the image conversion processing including normalization of the contour shape corresponding to the tuyere is performed using the fitting result, the accuracy of normalization is improved. As a result, in the blast furnace tuyere state observation apparatus 10 according to the present embodiment, it is possible to improve the calculation accuracy of the index representing the state in the tuyere. Thereby, the blast furnace tuyere state observation apparatus 10 according to the present embodiment can more accurately observe the state in the vicinity of the blast furnace tuyere without depending on fluctuations in the blast furnace tuyere image due to the heat.

  Moreover, in the blast furnace tuyere state observation apparatus 10 according to the present embodiment, each of the two states, that is, the loss of raw minerals and the expansion of pulverized coal, can be quantitatively determined independently of each other. In the state where raw ore dropping and pulverized coal expansion occurred, characteristic bright part distribution information and trajectory information are observed compared to the case where the state of the blast furnace tuyere is good, so the occurrence of these phenomena, Judgment can be easily made without relying on a sensory test. Moreover, in the blast furnace tuyere state observation apparatus 10 according to the present embodiment, since the number of feature quantities used for processing is two, it can be easily visualized by a locus on the mn plane.

(About blast furnace tuyere state observation method)
Next, an example of the flow of the blast furnace tuyere state observation method performed by the blast furnace tuyere state observation apparatus according to the embodiment of the present invention will be briefly described with reference to FIGS. 19 and 20. FIG. 19 is a flowchart illustrating an example of the flow of the blast furnace tuyere state observation method according to the present embodiment, and FIG. 20 is a flowchart illustrating an example of the flow of the elliptical shape specifying method according to the present embodiment.

<Overall flow of blast furnace tuyere state observation method>
The imaging device 100 of the blast furnace tuyere state observation device 10 according to the present embodiment images the tuyere under the control of the imaging control unit 201 in the arithmetic processing device 200, and generates a thermal radiance image (step S101). ), And outputs the generated thermal radiance image to the arithmetic processing unit 200.

  When the image processing unit 203 included in the arithmetic processing unit 200 of the blast furnace tuyere state observation apparatus 10 acquires the thermal radiance image output from the imaging apparatus 100, the acquired thermal radiance image data is converted into the elliptical shape specifying unit 211. Transmit to. The elliptical shape specifying unit 211 uses the transmitted thermal radiance image to specify an elliptical shape that matches the contour shape of the tuyere (step S103). The elliptical shape specifying unit 211 transmits information representing the elliptical parameters of the specified elliptical shape to the image converting unit 213.

  The image conversion unit 213 performs geometric conversion on the thermal radiance image using the elliptical shape specified by the elliptical shape specification unit 211, and generates a normalized image (step S105). Subsequently, the image conversion unit 213 performs polar coordinate conversion on the generated normalized image (step S107), and the normalized image (band image) after the polar coordinate conversion is converted into a binarized image included in the index calculation unit 215. The data is output to the generation unit 217.

  The binarized image generation unit 217 binarizes the normalized image (that is, the band-shaped image) after the polar coordinate conversion based on the binarization threshold value to generate a binarized image (step S109), and the generated binary The converted image is output to the bright part distribution information generation unit 219.

  The bright part distribution information generation unit 219 generates bright part distribution information based on the binarized image output from the binarized image generation unit 217 (step S111), and uses the generated bright part distribution information as a feature amount calculation unit. To 221.

  The feature quantity calculation unit 221 refers to the bright part distribution information output from the bright part distribution information generation unit 219, and the number-of-elements feature quantity and the radial position feature quantity (for example, the bright part average pixel number m and the maximum pixel radius). n) is calculated (step S113). Thereafter, the feature amount calculation unit 221 outputs the calculated feature amounts to the trajectory information generation unit 223.

  The trajectory information generation unit 223 associates a point defined by the combination of feature amounts calculated in the feature amount coordinate system based on the feature amount calculated by the feature amount calculation unit 221, and the feature amount corresponding to the time transition Trajectory information indicating the change in the number is generated (step S115). Thereafter, the trajectory information generation unit 223 outputs the generated trajectory information to the state determination unit 225.

  The state determination unit 225 determines the state of the blast furnace tuyere based on the time transition of the trajectory information generated by the trajectory information generation unit 223 (step S117).

  Heretofore, an example of the overall flow of the blast furnace tuyere state observation method according to the present embodiment has been briefly described with reference to FIG.

<Flow of ellipse shape identification method>
Next, an example of the flow of the elliptic shape specifying method performed by the elliptic shape specifying unit 211 will be briefly described with reference to FIG.

  When the saturation processing unit 231 included in the elliptical shape specifying unit 211 acquires the thermal radiance image output from the imaging device 100, the saturation processing unit 231 performs a saturation process or a binarization process on the acquired thermal radiance image (Step S21). S121). Thereafter, the saturation processing unit 231 outputs the obtained thermal radiance image after the saturation processing to the contour extraction unit 233.

  The contour extracting unit 233 extracts the tuyere-shaped contour of the thermal radiance image after the saturation processing output from the saturation processing unit 231 by the method illustrated in FIG. 8 to generate a contour strength image (step S123). ). After that, the contour extraction unit 233 outputs the generated contour strength image to the shape calculation unit 235.

  When the shape calculation unit 235 acquires the contour strength image from the contour extraction unit 233, the shape calculation unit 235 sets initial values of the elliptical parameters used for the elliptical fitting process to be performed (step S125). More specifically, when the elliptical fitting process to be performed is n = 1st processing, or n = Nth processing, and n = N−1th fitting processing has failed. In this case, the shape calculation unit 235 sets the system initial value registered as the system parameter in the arithmetic processing unit 203 as the initial value of the elliptical parameter. On the other hand, if n = N-th processing and n = N−1-th fitting processing is successful, the processing is preset based on the elliptical shape calculated by the n = N−1-th fitting processing. The set parameter is set as the initial value of the elliptical parameter.

  Next, the shape calculation unit 235 “identifies the elliptical shape parameter that gives the maximum value of the evaluation function E defined by the formula 101” while changing the set elliptical shape parameter in the direction in which the ellipse shrinks. The elliptical fitting process for obtaining the solution of the optimization problem is performed using the gradient method (step S127).

  At this time, the shape calculation unit 235 determines whether or not the optimization calculation has converged and the elliptical fitting process has been successful (step S129). If the elliptical fitting process is successful, the shape calculating unit 235 outputs the specified elliptical parameters and uses the specified elliptical parameters as the initial elliptical parameter values for the next and subsequent times. Set (step S131). On the other hand, if the elliptical fitting process fails without converging, the shape calculation unit 235 sets the system initial value as an elliptical parameter (step S133), and the elliptical fitting process in step S127 is performed again. carry out.

  The processing described above is performed by each processing unit included in the elliptical shape specifying unit 211, whereby an elliptical shape corresponding to the tuyere contour of the thermal radiance image is specified.

  The example of the flow of the blast furnace tuyere state observation method implemented by the blast furnace tuyere state observation apparatus according to the embodiment of the present invention has been briefly described above with reference to FIGS. 19 and 20.

(About hardware configuration)
Next, the hardware configuration of the arithmetic processing device 200 according to the embodiment of the present invention will be described in detail with reference to FIG. FIG. 21 is a block diagram for explaining a hardware configuration of the arithmetic processing device 200 according to the embodiment of the present invention.

  The arithmetic processing device 200 mainly includes a CPU 901, a ROM 903, and a RAM 905. The arithmetic processing device 200 further includes a bus 907, an input device 909, an output device 911, a storage device 913, a drive 915, a connection port 917, and a communication device 919.

  The CPU 901 functions as an arithmetic processing device and a control device, and controls all or a part of the operation in the arithmetic processing device 200 according to various programs recorded in the ROM 903, the RAM 905, the storage device 913, or the removable recording medium 921. The ROM 903 stores programs used by the CPU 901, calculation parameters, and the like. The RAM 905 primarily stores programs used by the CPU 901, parameters that change as appropriate during execution of the programs, and the like. These are connected to each other by a bus 907 constituted by an internal bus such as a CPU bus.

  The bus 907 is connected to an external bus such as a PCI (Peripheral Component Interconnect / Interface) bus via a bridge.

  The input device 909 is an operation unit operated by the user, such as a mouse, a keyboard, a touch panel, a button, a switch, and a lever. The input device 909 may be, for example, remote control means (so-called remote control) using infrared rays or other radio waves, or may be an external connection device 923 such as a PDA corresponding to the operation of the arithmetic processing device 200. May be. Furthermore, the input device 909 includes, for example, an input control circuit that generates an input signal based on information input by a user using the operation unit and outputs the input signal to the CPU 901. The user of the arithmetic processing device 200 can input various data and instruct processing operations to the arithmetic processing device 200 by operating the input device 909.

  The output device 911 is configured by a device that can notify the user of the acquired information visually or audibly. Examples of such devices include CRT display devices, liquid crystal display devices, plasma display devices, EL display devices and display devices such as lamps, audio output devices such as speakers and headphones, printer devices, mobile phones, and facsimiles. The output device 911 outputs results obtained by various processes performed by the arithmetic processing device 200, for example. Specifically, the display device displays results obtained by various processes performed by the arithmetic processing device 200 as text or images. On the other hand, the audio output device converts an audio signal composed of reproduced audio data, acoustic data, and the like into an analog signal and outputs the analog signal.

  The storage device 913 is a data storage device configured as an example of a storage unit of the arithmetic processing device 200. The storage device 913 includes, for example, a magnetic storage device such as an HDD (Hard Disk Drive), a semiconductor storage device, an optical storage device, or a magneto-optical storage device. The storage device 913 stores programs executed by the CPU 901, various data, various data acquired from the outside, and the like.

  The drive 915 is a recording medium reader / writer, and is built in or externally attached to the arithmetic processing unit 200. The drive 915 reads information recorded on a removable recording medium 921 such as a mounted magnetic disk, optical disk, magneto-optical disk, or semiconductor memory, and outputs the information to the RAM 905. The drive 915 can also write a record on a removable recording medium 921 such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory. The removable recording medium 921 is, for example, a CD medium, a DVD medium, a Blu-ray medium, or the like. The removable recording medium 921 may be a CompactFlash (registered trademark) (CompactFlash: CF), a flash memory, an SD memory card (Secure Digital memory card), or the like. Further, the removable recording medium 921 may be, for example, an IC card (Integrated Circuit card) on which a non-contact IC chip is mounted, an electronic device, or the like.

  The connection port 917 is a port for directly connecting a device to the arithmetic processing device 200. Examples of the connection port 917 include a USB (Universal Serial Bus) port, an IEEE 1394 port, a SCSI (Small Computer System Interface) port, and an RS-232C port. By connecting the external connection device 923 to the connection port 917, the arithmetic processing apparatus 200 acquires various data directly from the external connection device 923 or provides various data to the external connection device 923.

  The communication device 919 is a communication interface configured with, for example, a communication device for connecting to the communication network 925. The communication device 919 is, for example, a communication card for wired or wireless LAN (Local Area Network), Bluetooth (registered trademark), or WUSB (Wireless USB). The communication device 919 may be a router for optical communication, a router for ADSL (Asymmetric Digital Subscriber Line), or a modem for various communication. The communication device 919 can transmit and receive signals and the like according to a predetermined protocol such as TCP / IP, for example, with the Internet and other communication devices. The communication network 925 connected to the communication device 919 is configured by a wired or wireless network, and may be, for example, the Internet, a home LAN, infrared communication, radio wave communication, satellite communication, or the like. .

  Heretofore, an example of the hardware configuration capable of realizing the function of the arithmetic processing device 200 according to the embodiment of the present invention has been shown. Each component described above may be configured using a general-purpose member, or may be configured by hardware specialized for the function of each component. Therefore, it is possible to change the hardware configuration to be used as appropriate according to the technical level at the time of carrying out this embodiment.

  Hereinafter, the blast furnace tuyere state observation method and the blast furnace tuyere state observation apparatus according to the embodiment of the present invention will be specifically described with reference to experimental examples. The experimental example shown below is merely an example of the blast furnace tuyere state observation method and blast furnace tuyere state observation apparatus according to the embodiment of the present invention, and the blast furnace tuyere state observation method and blast furnace tuyere according to the present invention. The state observation apparatus is not limited to the following example.

  In the following, using a moving image of 3600 frames of width (X direction) 480 pixels × height (Y direction) 360 pixels as a thermal radiance image, an elliptical fitting process is performed in accordance with the method described above. Carried out. Here, the saturation processing unit 231 performs saturation processing with a saturation threshold of 30 and sets the parameter σ = 7 of the Gaussian filter used in the contour extraction unit 233. In addition, in the evaluation function E of the expression 101 used in the shape calculation unit 235, the optimization coefficient K is set to 20 and the initial value enlargement ratio in the next frame when the fitting process is successful is set to 1.05. .

  Here, the radius of the normalized circle generated by the image conversion unit 213 is set to 170 pixels in consideration of the size of the image frame used for the processing.

  In the following, focusing on the normalized circle generated by the elliptical shape specifying unit 211 and the image converting unit 213 in which the above parameters are set, the contour position on the X axis positive direction side of the normalized circle shown in FIG. 22A A histogram of the X coordinate of the point A is calculated. Since the center X coordinate of the thermal radiance image = 480 ÷ 2 = 240, the image conversion unit 213 performs image conversion so that the coordinates of the point C are (X, Y) = (240, 180). Thus, the X coordinate target of point A is 240 + 170 = 410.

  The obtained result is shown in FIG. 22B. In FIG. 22B, “the input image remains” means that a fixed ellipse obtained in advance according to the method disclosed in Patent Document 1 without performing the elliptical shape specifying process according to the embodiment of the present invention. Is applied to the tuyere image, and the degree of variation in the X coordinate of point A when a normalized circle is generated from the resulting ellipse is shown. Further, in FIG. 22B, “elliptical shape specification” indicates the degree of variation in the X coordinate of the point A when the normalized circle is generated after performing the elliptical shape specification processing according to the embodiment of the present invention. Is.

  As is apparent from FIG. 22B, when the elliptical shape specifying process according to the embodiment of the present invention is not performed, the variation in the coordinates of the point A is large, and the pixel coordinates are also 410 pixels, which is the target value. It was very different. On the other hand, when the elliptical shape specifying process according to the embodiment of the present invention is performed, the pixel coordinates of the point A are within a very narrow range centering around the target value of 410 pixels, This shows that the resulting image variation is suppressed.

  Next, according to the embodiment of the present invention, a total of 10 cases of five moving images in which it is determined by visual observation that pulverized coal expansion has occurred and five moving images in which it has been determined that raw mining has occurred have occurred. The state in the blast furnace was determined using the observation method including the ellipse shape specifying process and the observation method disclosed in Patent Document 1 that does not perform the ellipse shape specifying process according to the embodiment of the present invention. . In the following processing, in the feature amount space defined by the average number of bright area pixels and the maximum pixel radius, pulverized coal expansion occurs in an area where the average number of bright area pixels ≦ 100 and the maximum pixel radius ≧ 108. The area was determined to be.

  Below, the observation method disclosed in Patent Document 1 is taken as a conventional example, and the obtained results are shown in Table 1 and FIG. 23A below, and the observation method according to the embodiment of the present invention is obtained as an example of the present invention. The results are shown in Table 2 below and FIG. 23B. In FIG. 23A and FIG. 23B, the area | region represented by the oblique line in a graph corresponds to the area | region where it is judged that said pulverized coal expansion has generate | occur | produced.

  As is clear from Tables 1 and 2 and FIGS. 23A and 23B, in the conventional example based on the observation method disclosed in Patent Document 1, it is determined that 2 cases out of 5 cases of moving images of pulverized coal expansion are raw ore missing. However, in the example of the present invention, all of the moving images of the pulverized coal expansion in the five cases were determined to be pulverized coal expansion.

  As described above, by using the blast furnace tuyere state observation method according to the embodiment of the present invention, it is possible to observe the state near the blast furnace tuyere more accurately without depending on the fluctuation of the blast furnace tuyere image due to the heat. It became clear that.

  The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings, but the present invention is not limited to such examples. It is obvious that a person having ordinary knowledge in the technical field to which the present invention pertains can come up with various changes or modifications within the scope of the technical idea described in the claims. Of course, it is understood that these also belong to the technical scope of the present invention.

DESCRIPTION OF SYMBOLS 10 Blast furnace tuyere state observation apparatus 100 Imaging apparatus 200 Arithmetic processing apparatus 201 Imaging control part 203 Image processing part 205 Display control part 207 Storage part 211 Ellipse shape specific part 213 Image conversion part 215 Index calculation part 217 Binary image generation part 219 Bright part distribution information generation unit 221 Feature amount calculation unit 223 Trajectory information generation unit 225 State determination unit 231 Saturation processing unit 233 Contour extraction unit 235 Shape calculation unit

Claims (11)

A method for observing the state in the tuyere using a thermal radiance image captured by an imaging device provided in the blast furnace tuyere,
Applying an elliptical shape to the contour shape of the tuyere in each acquired thermal radiance image, an elliptical shape identifying step for identifying an elliptical shape that matches the contour shape;
Using the ellipse parameters of the ellipse shape identified in the ellipse shape identification step, each of the thermal radiance images is subjected to a normalization circle having a constant center position and a constant radius. An image conversion step for performing geometric transformation to generate a normalized image and performing polar coordinate transformation on the normalized image;
A binarized image generating step for binarizing the normalized image generated in the image converting step and generating the binarized image;
A bright part distribution information generating step for generating bright part distribution information indicating a distribution in a radial direction of the normalized circle of bright parts existing in the binarized image using the generated binary image; ,
A method for observing blast furnace tuyere status.   The ellipse parameters of the specified ellipse shape are the center position coordinates of the ellipse shape, the long axis direction length of the ellipse shape corresponding to the image horizontal direction of the thermal radiance image, and the image vertical direction of the thermal radiance image. 2. The blast furnace tuyere state observation method according to claim 1, wherein the length is an elliptical short-axis direction length corresponding to.   In the ellipse shape specifying step, an evaluation function including a line integral of the contour strength located on the arc of the elliptical shape to be applied is defined for the contour strength image in which the image luminance value of the contour portion of the tuyere is emphasized, and the evaluation The blast furnace tuyere state observation method according to claim 1 or 2, wherein an ellipse shape that matches the contour shape is identified by solving an optimization problem that identifies an ellipse shape that gives a local maximum value of the function.   In the ellipse shape specifying step, a value obtained by subtracting the arc length of the ellipse from a constant multiple of the line integral of the contour intensity is used as the evaluation function, and an elliptical shape surrounding the outside of the tuyere is used as the initial value of the elliptical shape. The blast furnace tuyere state observation method according to claim 3, wherein an optimization process is performed to identify an elliptical shape that is set and gives the maximum value by a gradient method.   In the elliptical shape specifying step, two types of differential images are generated by differentiating the thermal radiance image in the horizontal direction and the vertical direction of the image, respectively, and the two types of differential images are squared and added together. The blast furnace tuyere state observation method according to any one of claims 1 to 4, wherein a contour strength image is generated and the elliptical shape is identified based on the contour strength image.   In the elliptical shape specifying step, prior to the generation of the differential image, the thermal radiance image is subjected to a two-dimensional Gaussian filter process, and the thermal radiance image after the two-dimensional Gaussian filter process is used, The blast furnace tuyere state observation method according to claim 5, wherein a differential image is generated.   7. The contour intensity image is generated according to any one of claims 5 and 6, wherein in the elliptical shape specifying step, a saturation process or a binarization process is performed on the thermal radiance image. Blast furnace tuyere state observation method. In the elliptical shape specifying step,
The initial value of the optimization process for the first thermal radiance image of each of the thermal radiance images, and the initial value of the next optimization process when the optimization process fails are defined in advance. Oval shape,
In the optimization process for the first thermal radiance image and the optimization process other than the next optimization process when the optimization process fails, the long axis of the elliptical shape that is the solution of the previous optimization process The blast furnace tuyere state observation method according to any one of claims 3 to 7, wherein an initial value is obtained by multiplying the length in the direction and the length in the minor axis direction by a constant number.   In the bright part distribution information generation step, pixel values corresponding to a plurality of pixels having the same radial position and different declination positions are provided for each radial position in the binarized image. The blast furnace tuyere state observation method according to any one of claims 1 to 8, wherein integration is performed and the obtained integrated value is used as an element of the bright portion distribution information at the corresponding radial position. A feature amount calculating step for calculating a feature amount characterizing the bright portion distribution information at a certain time using the generated bright portion distribution information;
Using the calculated feature quantity, a feature quantity coordinate system defined using the feature quantity is set, and specified by a combination of the feature quantities corresponding to the bright part distribution information at a plurality of different times. A trajectory information generation step for generating trajectory information indicating a trajectory associated with a time transition of a point in the feature amount coordinate system;
A state determination step of determining a state in the tuyere using the generated trajectory information;
Further including
In the feature amount calculating step, at least an element number feature amount characterizing the number of elements included in a predetermined radial direction range, and a radial position feature amount indicating the radial position giving the maximum value of the element are at least The blast furnace tuyere state observation method according to claim 9, which is calculated. An imaging device provided at a blast furnace tuyere and capturing a thermal radiance image at the blast furnace tuyere,
An arithmetic processing device that performs image processing on each of the thermal radiance images captured by the imaging device;
With
The arithmetic processing unit includes:
Applying an elliptical shape to the contour shape of the tuyere in each acquired thermal radiance image, an elliptical shape identifying unit that identifies an elliptical shape that matches the contour shape;
Using each of the elliptical parameters of the elliptical shape specified by the elliptical shape specifying unit, each of the contour shapes is a normalized circle having a constant center position and a constant radius. An image conversion unit that performs geometric conversion to generate a normalized image and performs polar coordinate conversion on the normalized image;
A binarized image generating unit that binarizes the normalized image generated by the image converting unit after the polar coordinate conversion and generates a binarized image;
A bright part distribution information generating unit that generates bright part distribution information indicating a distribution in a radial direction of the normalized circle of bright parts existing in the binarized image using the generated binary image; ,
A blast furnace tuyere state observation device.