JP7003031B2 - Melting furnace monitoring device, melting furnace monitoring method, and program - Google Patents

Description

The present invention relates to a melting furnace monitoring device, a melting furnace monitoring method, and a program.

In Patent Document 1, a thermal image of the sprue is taken by a thermal image camera installed diagonally below the sprue of the waste melting furnace, and the blockage of the sprue is monitored by the degree of decrease in the area of the high-brightness portion in the thermal image. It is stated that.

Japanese Patent No. 3242315

By the way, if the slag adheres to the slag opening for discharging the slag, the substantially opening ratio of the slag opening may decrease, and the discharge of the molten slag may be hindered. However, the brightness of the molten slag and the sticking flag are close to each other, and it is difficult to distinguish them.

The present invention has been made in view of the above problems, and a main object thereof is to provide a melting furnace monitoring device, a melting furnace monitoring method, and a program capable of discriminating stuck slag at a slag opening. It is in.

In order to solve the above problems, the melting furnace monitoring device according to one aspect of the present invention includes an image acquisition means for acquiring a plurality of images in which the outlet of the melting furnace is continuously photographed, and each of the plurality of images. The slag region extraction means for extracting the slag region by determination based on the brightness threshold, and the fixed slag region specifying means for specifying the region from which the slag region is extracted in common in the plurality of images as the fixed slag region are provided. ..

Further, in the melting furnace monitoring method of another aspect of the present invention, a plurality of images obtained by continuously photographing the outlet of the melting furnace are acquired, and in each of the plurality of images, a slag region is determined by a determination based on a luminance threshold. Is extracted, and the region from which the slag region is extracted in common in the plurality of images is specified as the fixed slag region.

Further, the program of another aspect of the present invention is an image acquisition means for acquiring a plurality of images in which the outlet of a melting furnace is continuously photographed, and in each of the plurality of images, a slag region is determined by a determination based on a brightness threshold. The computer functions as a slag region extraction means for extracting the slag region and a fixed slag region specifying means for specifying the region from which the slag region is extracted in common in the plurality of images as the fixed slag region.

According to the present invention, it is possible to discriminate the fixed slag at the outlet.

It is a figure which shows typically the melting furnace monitoring device and the gasification melting furnace. It is a block diagram which shows the structural example of the melting furnace monitoring apparatus. It is a flow chart which shows the procedure example of the melting furnace monitoring method. It is a flow chart which shows the procedure example of the slag area extraction processing. It is a figure which shows the example of the original image and the slag mouth shape template. It is a figure explaining the slag area extraction process. It is a figure which shows the example of the slag area image. It is a figure explaining the identification of the sticking slag region and the calculation of the slag opening ratio. It is a flow chart which shows the procedure example of the melting furnace monitoring method which concerns on the modification. It is a figure for demonstrating an example of an offset. It is a figure for demonstrating an example of a discriminant function. It is a figure for demonstrating an example of a detection result.

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings. It should be noted that each embodiment shown below exemplifies a method and an apparatus for embodying the technical idea of the present invention, and the technical idea of the present invention is not limited to the following. do not have. The technical idea of the present invention may be modified in various ways within the technical scope described in the claims.

FIG. 1 is a diagram schematically showing a melting furnace monitoring device 1 and a gasification melting furnace 2. The gasification melting furnace 2 is, for example, a gasification melting furnace used in a waste treatment facility. The gasification melting furnace 2 includes a melting furnace 21. The combustible gas thermally decomposed in the gasification furnace (not shown) and the unburned component flow into the furnace of the melting furnace 21 through the inflow path 23. The melting furnace 21 burns the unburned component at a high temperature and melts the ash component to generate slag. The molten slag is discharged from the slag opening 25 formed in the lower part of the melting furnace 21, and is taken out of the melting furnace 21 through the discharge path 27.

Below the melting furnace 21, a camera 3 for photographing the slag opening 25 is provided. The camera 3 is arranged diagonally below the slag opening 25, and takes a picture of the slag opening 25 through the monitoring hole 29 provided in the discharge path 27. The camera 3 outputs the photographed data obtained by photographing the slag opening 25 to the melting furnace monitoring device 1.

The melting furnace monitoring device 1 calculates the opening ratio of the slag opening 25 based on the photographing data supplied from the camera 3. The calculated aperture ratio information is presented to the worker P or transmitted to the control device of the gasification melting furnace 2. The worker P who has referred to the aperture ratio information or the received control device performs various operations so as to improve the aperture ratio of the slag opening 25.

FIG. 2 is a block diagram showing a configuration example of the melting furnace monitoring device 1. The melting furnace monitoring device 1 is realized by, for example, a server computer or the like. The melting furnace monitoring device 1 includes a control unit 10, a storage unit 12, an operation unit 13, a display unit 14, and an input / output unit 15.

The control unit 10 includes at least one microprocessor (CPU), and executes information processing according to a program stored in the storage unit 22. The storage unit 12 includes a main storage unit (for example, RAM) and an auxiliary storage unit (for example, a hard disk drive or a solid state drive).

The program is supplied to the storage unit 12 via an information storage medium such as an optical disk or a memory card. Alternatively, the program may be supplied to the storage unit 12 via a communication network such as the Internet.

The operation unit 13 is, for example, a keyboard, a mouse, a touch panel, or the like. The display unit 14 is, for example, a liquid crystal display, an organic EL display, or the like. The input / output unit 15 is an interface for connecting an external device such as a camera 3.

FIG. 3 is a flow chart showing a procedure example of the melting furnace monitoring method. FIG. 4 is a flow chart showing a procedure example of the slag region extraction process. The control unit 10 of the melting furnace monitoring device 1 executes the processes shown in FIGS. 3 and 4 according to the program.

First, the control unit 10 acquires a camera image taken by the camera 3 (S1: processing as an image acquisition means). The control unit 10 extracts a camera image every predetermined time (for example, every second) from the shooting data of the camera 3. The shooting data is, for example, moving image data or image data, and includes a plurality of camera images in which the exit 25 is continuously shot.

Next, the control unit 10 detects the position of the slag opening 25 on the camera image (S2). The control unit 10 uses a slag opening shape template (see FIG. 5A) corresponding to the shape and size of the slag opening 25, and performs template matching between the camera image and the slag opening shape template. The position of the slag opening 25 on the camera image is specified (see FIG. 5B), and the control unit 10 moves the camera image in parallel so that the slag opening 25 comes to the center of the image.

In template matching, for example, a position where the correlation coefficient between the camera image and the exit shape template is maximized is searched for. Not limited to this, for example, the position where the cross-correlation function is maximized may be searched, the position where the square difference is minimized may be searched, or the position where the multiplication value is maximized may be searched. ..

Next, the control unit 10 executes a luminance unevenness removing process for suppressing the luminance unevenness of the camera image (S3: processing as a luminance unevenness suppressing means). The control unit 10 suppresses uneven brightness of the camera image by applying a filter (for example, a low-cut filter) that limits components having a frequency of a predetermined value or less to the camera image. Not limited to this, for example, a bandpass filter may be applied to remove fine noise as well as uneven luminance.

Since the camera image of the slag opening 25 uses the combustion reaction in the furnace as a light source, the brightness may be non-uniform depending on the combustion state. For example, an image having uneven brightness is often taken, in which the center of the image is the brightest and the image gradually becomes darker toward the periphery (see FIG. 6A). In such a camera image, the slag may be brighter than the furnace wall at the back of the slag opening 25, so if the binarization process is applied as it is, the slag region may not be extracted well.

Therefore, in the present embodiment, an image after removing unevenness (FIG. 6B) in which uneven brightness of the camera image is suppressed is created, and a slag region is extracted from the image after removing unevenness. This makes it possible to detect the slag region without being affected by the uneven brightness even if the camera image has uneven brightness.

Next, the control unit 10 executes a slag region extraction process (S4: process as a slag region extraction means). In the slag region extraction process, the control unit 10 applies a binarization process to the camera image in which the luminance unevenness is suppressed, and extracts the region whose luminance is equal to or less than the threshold value as the slag region.

An example of the procedure for the slag region extraction process shown in FIG. 4 will be described. First, the control unit 10 extracts a sudden change in brightness in the camera image (S41: Processing as a means for extracting the sudden change in brightness, FIG. 6 (c)). Extraction of the sudden change in brightness is performed by applying edge enhancement processing and binarization processing to the camera image.

Examples of the edge enhancement process include a Gradient filter, a Sobel filter, a Laplace filter, a Canny method, and a high-pass filter. By this processing, an edge-enhanced image having a large pixel value can be obtained in the sudden change in brightness. Examples of the binarization process include Otsu's binarization method (discriminant analysis method), P-tile method, and mode method. By this process, the threshold value is automatically determined from the luminance histogram, and the edge-enhanced image is binarized. As a result of these processes, the sudden change in brightness is extracted. In addition to these processes, an expansion filter may be used to expand the sudden change in luminance.

Next, the control unit 10 calculates the brightness histogram by removing the sudden change in brightness from the camera image (S42). The control unit 10 creates a brightness sudden change part removal mask (FIG. 6 (d)) for removing the brightness sudden change part, and is an effective area of the brightness sudden change part removal mask in the image after unevenness removal (FIG. 6 (b)). The luminance histogram (FIG. 6 (e)) is calculated from the region corresponding to the (white portion). The brightness sudden change part removal mask is created based on the extraction result of the brightness sudden change part and the slag opening shape template. The area other than the sudden change part is regarded as the effective area (white part).

The reason for excluding the sudden change in brightness will be explained. In the image after removing the unevenness in which the uneven brightness of the camera image is suppressed, the sudden change in brightness may become high brightness (see FIG. 6B). If the threshold value for the binarization process is automatically determined in such an image after removing unevenness, the threshold value is determined so as to separate the sudden change in brightness from the rest, and there is a possibility that the slag region cannot be extracted well.

Therefore, in the present embodiment, the luminance histogram is calculated by removing the sudden change in luminance from the image after removing the unevenness, and the threshold value for the binarization process is determined. This makes it possible to extract the slag region from the image after removing the unevenness even if the sudden change in brightness becomes high in the image after removing the unevenness. The process (S41 and S42) for removing the sudden change in brightness is not essential and may be omitted.

Next, the control unit 10 automatically determines the threshold value based on the luminance histogram and executes the binarization process (S43). The control unit 10 determines and determines the threshold value having the highest separability from the luminance histogram (FIG. 6 (e)) by, for example, Otsu's binarization method (discrimination analysis method), P tile method, mode method, or the like. A binarization process is performed using the obtained threshold value to obtain a binarized image (FIG. 6 (f)). Then, the control unit 10 obtains a slag region image (FIG. 6 (h)) by taking a difference between the slag opening shape template (FIG. 6 (g)) and the binarized image (FIG. 6 (f)). obtain. The threshold may be manually determined by the worker.

The slag region image obtained in this way has a slag region (white portion) representing the slag at the slag opening 25 and a non-slag region (black-painted portion) representing other areas (for example, the furnace wall at the back of the slag opening 25). ) And (see FIG. 7).

Returning to the description of FIG. When the slag region extraction process (S4) is completed, the control unit 10 specifies a region from which the slag region has been extracted in common for a certain period as a fixed slag region (S5: processing as a fixed slag region specifying means).

Specifically, the control unit 10 generates a plurality of slag region images (FIG. 8 (h)) from a plurality of camera images acquired at predetermined time intervals (for example, every 1 second) acquired in a fixed period, and these are generated. A region in which the slag region is extracted in common in the slag region image is specified as a fixed slag region (FIG. 8 (i)), and a fixed slag image (FIG. 8 (j)) is obtained. The fixed slag image includes a fixed slag region (white portion) representing the fixed slag at the slag opening 25. Here, in the plurality of slag region images, the region where the presence or absence of the slag region changes with time is determined as molten slag, and the region where the presence or absence of the slag region does not change with time is determined as fixed slag.

Next, the control unit 10 calculates the aperture ratio of the slag opening 25 from the fixed slag region and the position / shape of the slag opening 25 (S6: processing as an opening ratio calculation means).

Specifically, the control unit 10 takes a difference between the outlet shape template (FIG. 8 (g)) and the fixed slag image (FIG. 8 (j)) to obtain an opening image (FIG. 8 (k)). ). The opening image includes an opening region (white portion) representing a substantial opening of the slag opening 25. Then, the control unit 10 calculates the aperture ratio of the slag opening 25 from the opening image (FIG. 8 (k)) and the slag opening shape template (FIG. 8 (g)) (FIG. 8 (m)). The aperture ratio of the outlet 25 is calculated as the ratio of the area of the opening region in the opening image to the area of the outlet 25 in the outlet shape template.

Finally, the control unit 10 presents the calculated opening ratio of the slag opening 25 to the operator (S7), and ends the process. The presentation to the operator is performed, for example, by displaying the opening ratio of the slag opening 25 on the display unit 14.

In the embodiment described above, by specifying a region that is commonly extracted in a plurality of continuously captured camera images as a slag region, it is fixed to molten slag that is difficult to distinguish from a single camera image alone. The slag is discriminated. This makes it possible to specify the fixed slag region at the slag opening 25 and accurately calculate the substantial opening ratio of the slag opening 25.

Hereinafter, a modified example will be described.

In the above embodiment, in S43 of FIG. 4, the threshold value for the binarization process is determined from the luminance histogram by a method such as Otsu's binarization method (discriminant analysis method). As shown in FIG. 6 (e), this is a distribution caused by a slag having a relatively low luminance (hereinafter referred to as a slag portion) and a distribution caused by an aperture having a relatively high luminance (hereinafter referred to as an aperture) on the luminance histogram. This is because it is separated from the part) to some extent.

However, depending on the amount and position of the slag and the operating condition of the furnace, for example, when the amount of slag is small, the slag portion and the opening may not be separated on the luminance histogram and may become monomodal. In that case, there is a possibility that the threshold value for separating the slag portion and the opening portion cannot be obtained well.

Therefore, in this modification, the threshold value for accurately separating the slag portion and the opening portion is determined by the procedure described below even if the slag portion and the opening portion are not separated on the luminance histogram. It is possible.

FIG. 9 is a flow chart showing a procedure example of the melting furnace monitoring method according to the modified example. The procedure that overlaps with the above embodiment is assigned the same number, and detailed description thereof will be omitted.

The control unit 10 calculates the luminance histogram from the camera image excluding the sudden change in luminance portion (S42), and then adds an offset value to a predetermined low luminance range to correct the luminance histogram (S53). Hereinafter, the luminance histogram corrected by this process is referred to as a "corrected luminance histogram".

The predetermined low luminance range is, for example, a section (including both ends) from the luminance value 0 to the image minimum luminance value. The control unit 10 adds an offset value to a part or all of the section from the luminance value 0 to the image minimum luminance value of the luminance histogram. Here, the slag portion side with respect to the opening of the luminance histogram is the low luminance side. Adding an offset value to the low brightness range of the luminance histogram corresponds to creating a slag portion on the luminance histogram in a pseudo manner.

Specifically, the offset value may be added only to the image minimum luminance value as shown in FIG. 10 (a), or the offset value is equal to the section from the luminance value 0 to the image minimum luminance value as shown in FIG. 10 (b). Or an offset value according to a normal distribution may be added to the section from the luminance value 0 to the image minimum luminance value as shown in FIG. 10 (c). The offset value may be, for example, the total frequency value of the entire luminance range (0 to 255) of the luminance histogram.

By using the corrected luminance histogram corrected in this way, even if the image contains only a small amount of slag and the slag portion and the opening portion are not separated on the luminance histogram, the slag portion and the opening portion are used. It is possible to obtain a threshold value that accurately separates and.

In this example, the offset value is added to the section from the luminance value 0 to the image minimum luminance value, but the present invention is not limited to this, and the offset value may be added to the interval slightly larger than the image minimum luminance value.

Returning to the description of FIG. 9, the control unit 10 calculates the discrimination function O (t) for each threshold candidate t based on the corrected luminance histogram (S54). The discriminant function O (t) is calculated every time the threshold candidate t of the luminance value is changed by, for example, 1 or 2 or more.

Here, the discriminant function O (t) is a function for discriminating between the low-luminance region and the high-luminance region by a statistical index based on the corrected luminance histogram, and in this example, it is for determining the threshold value of Otsu's binarization method. Use the function of. That is, a function for searching the maximum value of "maximum variance between classes / minimum variance within class" shown in the following formula 1 is used.

However, ω _U (t) represents the number of pixels in the region above the threshold candidate t, ω _L (t) represents the number of pixels in the region below the threshold candidate t, and m _U represents the luminance average in the region above the threshold candidate t. The value is represented, and _mL represents the brightness average value in the region below the threshold candidate t.

Next, the control unit 10 calculates, for each threshold candidate t, a matching ratio R (t) in which the luminance sudden change portion representing the slag boundary is included in the region boundary portion when the threshold candidate t is applied (S55).

As the brightness sudden change portion, the one extracted in S41 may be diverted or may be newly extracted. In addition to the extraction method described in S41 above, the sudden change in brightness may be extracted by calculating a differential image or a quadratic differential image and binarizing it with an arbitrary threshold value, or the differential image or secondary. It may be extracted by binarizing with the threshold value derived from the brightness value average and the brightness value standard deviation of the next derivative image.

The region boundary portion may be extracted by applying the threshold candidate t and actually performing the binarization process, or by searching for the pixel P (x, y) satisfying the following formula 2 or formula 3. May be done.

However, MIN (a1, a2, a3, ..., An) is a function that returns the minimum value of a1 to an, and MAX (a1, a2, a3, ..., An) is a function of a1 to an. A function that returns the maximum value.

The area extracted as the sudden change in brightness is defined as the area A, the area extracted as the area boundary is defined as the area B (t), and the common area C (t) between the area A and the area B (t) is derived. The match ratio R (t) is calculated for each threshold candidate t by the formula 4.

When performing this calculation, one or both of the region A and the region B (t) may be expanded.

Next, the control unit 10 calculates the discrimination function J (t) for each threshold candidate t (S56), and sets the threshold candidate t'that maximizes the discrimination function J (t) as the threshold, and creates two images after removing the luminance unevenness. Threshold (S57).

The discriminant function J (t) is a product of the discriminant function O (t) calculated in S54 and the match ratio R (t) calculated in S55, as shown in the following mathematical formula 5. ..

That is, the discriminant function J (t) searches for a threshold value in which the degree of separation between the low-luminance region (slag portion) and the high-luminance region (opening portion) is high, and the matching ratio between the sudden change in brightness and the region boundary portion is high. In order to satisfy these two conditions, the threshold candidate t'that maximizes the discriminant function J (t) is adopted as the threshold.

FIG. 11 is a diagram for explaining an example of the discriminant function. As shown in the figure, the discriminant function J (t) includes the match ratio R (t), so that the maximum value as a threshold value is shifted to the lower luminance side than the discriminant function O (t).

By using such a discriminant function J (t), even if the image contains only a small amount of slag and the slag portion and the opening portion are not separated on the luminance histogram, the slag portion and the opening portion are not separated. It is possible to obtain a threshold value that accurately separates and.

Specifically, it is possible to make the region boundary between the low-luminance region and the high-luminance region at the time of binarization more coincident with the slag boundary, and it is possible to prevent erroneous detection of the aperture region as the slag region. Is possible.

FIG. 12 is a diagram for explaining an example of the detection result. (A) represents the original image, (b) represents the image after removing unevenness, (c) represents the mask for removing the sudden change in brightness, and (d) represents the detection result (no offset) based on the binary value of Otsu. (E) Represents the detection result (with offset) by binarization of Otsu, and (f) represents the detection result by the discrimination function J (t).

According to this, (e) has less false detection than (d), and (f) has less false detection with the opening region as the slag region than (e) (that is, it looks closer to (a)). ) Is understood.

In the modified example described above, both the correction of the luminance histogram (S53) and the threshold value determination (S54 to S56) by the discrimination function J (t) are performed, but the present invention is not limited to this, and only one of them is performed. May be carried out. Further, the threshold value may be determined only by the matching ratio R (t).

1 melting furnace monitoring device, 10 control section, 12 storage section, 13 operation section, 14 display section, 15 input / output section, 2 gasification melting furnace, 21 melting furnace, 23 inflow path, 25 outlet, 27 discharge channel, 29 Monitoring hole, 3 cameras

Claims (11)

An image acquisition means for acquiring a plurality of images in which the outlet of the melting furnace is continuously photographed, and
A slag region extraction means for extracting a slag region by determination based on a luminance threshold in each of the plurality of images,
A fixed slag region specifying means for specifying a region from which the slag region has been extracted in common in the plurality of images as a fixed slag region,
A melting furnace monitoring device equipped with. Further provided with an opening ratio calculating means for calculating the opening ratio of the slag opening based on the area of the fixed slag region.
The melting furnace monitoring device according to claim 1. Further provided with a luminance unevenness suppressing means for suppressing the luminance unevenness of the image,
The slag region extraction means extracts the slag region in the image in which the luminance unevenness is suppressed.
The melting furnace monitoring device according to claim 1 or 2. The luminance unevenness suppressing means includes a filter that limits components having a frequency of a predetermined value or less.
The melting furnace monitoring device according to claim 3. The slag region extraction means determines the threshold value based on the luminance distribution of the image in which the luminance unevenness is suppressed.
The melting furnace monitoring device according to claim 3 or 4. Further provided with a luminance sudden change portion extraction means for extracting the luminance sudden change portion of the image,
The slag region extraction means determines the threshold value based on the luminance distribution of the image in which the luminance unevenness is suppressed and the luminance sudden change portion is removed.
The melting furnace monitoring device according to any one of claims 3 to 5. The slag region extraction means determines the threshold value based on a corrected luminance distribution obtained by adding an offset value to a predetermined low luminance range of the luminance distribution.
The melting furnace monitoring device according to claim 5 or 6. Further provided with a luminance sudden change portion extraction means for extracting the luminance sudden change portion of the image,
The slag region extraction means determines the threshold value based on the matching ratio between the region boundary portion between the slag region and the non-slag region and the brightness sudden change portion when the threshold value candidate is applied.
The melting furnace monitoring device according to any one of claims 5 to 7. The slag region extraction means is a threshold candidate having a maximum discriminant function obtained by multiplying a discriminant function representing the degree of separation between a low-luminance region serving as a slag region and a high-luminance region serving as a non-slag region with the matching ratio. Is determined as the threshold value ,
The melting furnace monitoring device according to claim 8. Acquired multiple images of the outlet of the melting furnace taken continuously,
In each of the plurality of images, a slag region is extracted by determination based on the luminance threshold value.
A region in which the slag region is extracted in common in the plurality of images is specified as a fixed slag region.
Melting furnace monitoring method. Image acquisition means for acquiring multiple images of continuous images of the outlet of a melting furnace,
A slag region extraction means for extracting a slag region by determination based on a luminance threshold in each of the plurality of images, and a slag region extraction means.
A means for specifying a fixed slag region, which specifies a region from which the slag region is extracted in common in the plurality of images as a fixed slag region.
A program that makes your computer work as.

Images