WO2012153649A1

WO2012153649A1 - Internal inspection method for glass-melting furnace, operation method for glass-melting furnance, and internal inspection system for glass-melting furnace

Info

Publication number: WO2012153649A1
Application number: PCT/JP2012/061252
Authority: WO
Inventors: 博信黒石; 鈴木　俊彦; 信楜澤; 亮介赤木
Original assignee: 旭硝子株式会社
Priority date: 2011-05-06
Filing date: 2012-04-26
Publication date: 2012-11-15
Also published as: KR20140015357A; KR101923239B1; CN103415476A; JPWO2012153649A1; JP5928451B2; CN103415476B; TWI522326B; TW201247577A

Abstract

　Provided is an internal inspection system for glass-melting furnaces that enables the continuous observation of a specific region inside a glass-melting furnace. An image comprising a reference pattern provided inside a glass-melting furnace and and a specific range of the liquid surface of a molten glass base material is recorded. A region corresponding to the specific range is extracted from the recorded image on the basis of the reference pattern pictured in the image. Then, a background image which is to act as the background of a batch pile is generated on the basis of a plurality of extracted images extracted from a plurality of images. Background exclusion images, in which the background is removed from the extracted images in which the batch pile and the background are pictured, are generated by carrying out a process on each pixel whereby the brightness value in the background image is subtracted from the brightness value of the corresponding pixel in the extracted image. Observation data relating to the batch pile is then calculated on the basis of the background exclusion images.

Description

Glass melting furnace monitoring method, glass melting furnace operating method, glass melting furnace monitoring system

The present invention relates to a glass melting furnace monitoring method, a glass melting furnace operating method, a glass melting furnace monitoring system, and a glass article manufacturing method.

In the glass manufacturing process, there is a process in which a glass raw material is charged into a glass melting furnace and the raw material is melted in the glass melting furnace. The raw material charged into the glass melting furnace is a solid and gradually melts in the glass melting furnace. The raw material that has been charged and accumulated in the glass melting furnace is called a batch pile. The batch pile gradually moves along the flow of the molten glass that is the molten raw material (that is, from the upstream side to the downstream side of the glass melting furnace). In addition, the batch piles are gradually dissolved because they are dissolved by heat. The behavior of the batch pile is a guideline for the operation of the glass melting furnace, so the batch pile in the glass melting furnace is visually observed from the observation window provided in the glass melting furnace or sketched. It was. When observing a batch mountain, the part above the surface (namely, liquid level) of a molten glass becomes an observation object.

In addition, various methods have been proposed for monitoring a batch mountain by arranging a camera in an observation window in a glass melting furnace without relying on visual observation or sketching.

For example, in the technique described in Non-Patent Document 1, the Hough transform capable of detecting a straight line is used for determining the monitoring area. Non-Patent Document 1 describes obtaining the occupancy ratio of a batch mountain.

Patent Document 1 describes that a batch mountain is photographed, and the position and shape of the boundary line between the batch mountain and the liquid surface or the most downstream position are compared at each photographing time.

Further, Patent Document 2 scans the liquid level in the furnace to capture an image, obtains a position-luminance characteristic line from the image, and determines the location of the batch mountain based on the position-luminance characteristic line. A method is described.

In addition, Patent Document 3 describes a method for measuring and adjusting parameters relating to a raw material melted in a glass melting furnace.

Also, as a basic method of extracting a specific object from an image, there is a method of binarizing pixels. There are various methods for binarization, for example, there is a method for identifying a valley of a histogram of a pixel according to a luminance value and dividing the pixel into two classes. As a method for specifying a valley of a histogram of a pixel corresponding to a luminance value, a mode method, a discriminant analysis binarization method, or the like is known. The mode method is described in Non-Patent Documents 2 and 3. The discriminant analysis binarization method is described in Non-Patent Document 3. In the discriminant analysis binarization method, when the histogram is divided into two classes, the threshold value is determined so that the separation between the two classes is the best. Specifically, a threshold value that maximizes the variance ratio between the intra-class variance and the inter-class variance for the background area and the specific object area in the image is determined.

Japanese Unexamined Patent Publication No. 2009-161396 Japanese Unexamined Patent Publication No. 59-44606 US Patent Application Publication No. 2004/0079113

When monitoring a batch pile by placing a camera in the observation window of the glass melting furnace, it is possible to continue to observe a certain area in the glass melting furnace and to accurately monitor the state of the batch pile in that area. preferable.

However, the camera position and orientation may shift during maintenance work such as cleaning the observation window. Then, the shooting range of the camera is also shifted. As described above, when the position and orientation of the camera change, the accuracy of evaluation of the time-dependent change of the batch mountain state is lowered.

In addition, bubbles are generated on the liquid surface of the dissolved raw material when the raw material is heated. Therefore, when a batch mountain in the glass melting furnace is photographed, an image of the batch mountain with a background of bubbles is obtained. In order to accurately monitor the state of the batch crest, it is preferable to separate the bubbles and the batch crest in the image and extract the batch crest portion from the image.

It is also preferable to operate the glass melting furnace by properly grasping which operating parameters of the glass melting furnace should be adjusted according to the monitoring result when the batch mountain is monitored.

Therefore, an object of the present invention is to provide a glass melting furnace monitoring method and a glass melting furnace monitoring system capable of satisfactorily continuing observation of a certain region in the glass melting furnace. Moreover, it aims at providing the manufacturing method of the glass article which manufactures a glass article, implement | achieving such a favorable observation state.

Also, an object of the present invention is to provide a glass melting furnace operating method capable of clarifying which operating parameter of the glass melting furnace should be adjusted according to the state of the monitored batch mountain.

In the monitoring method in the glass melting furnace according to the present invention, the image photographing means photographs an image including a reference pattern provided in the glass melting furnace and a certain range on the liquid surface of the glass raw material melted in the glass melting furnace. An image capturing step and an area extracting step for extracting an area corresponding to a certain range from the captured image according to the attitude of the image capturing means calculated using the position shift of the reference pattern captured in the image And a background image creation step of creating a background image as a background of a batch mountain that is a glass raw material accumulated in a glass melting furnace based on a plurality of extracted images extracted from a plurality of images as a region corresponding to a certain range And a process of subtracting the luminance value of the corresponding pixel in the background image from the luminance value of the pixel of the extracted image extracted as an area corresponding to a certain range from the photographed image. This is done for each element, and a background excluded image generation step that generates a background excluded image that excludes the background from the extracted image in which the batch mountain and the background are reflected, and observation data related to the batch mountain are calculated based on the background excluded image. And an observation data calculation step.

In the background image creation step, the number of pixels corresponding to each luminance value is counted for each corresponding pixel or corresponding area of the plurality of extracted images, and the background is determined based on the count result of the pixels corresponding to each luminance value. A method of creating a background image by determining the luminance value to be represented may be used.

In the background excluded image generation step, a process for subtracting the luminance value of the corresponding pixel in the background image from the luminance value of the pixel of the extracted image extracted as an area corresponding to a certain range from the captured image is performed for each pixel. A method of generating a background excluded image by binarizing each subtraction result may be used.

A background image conversion step for converting a background image into an image when a certain range is observed from above facing the liquid surface, and an extracted image extracted as an area corresponding to the certain range, the certain range facing the liquid surface An extracted image conversion step that converts the image into an image when observed from above, and in the background excluded image generation step, the background value after the conversion by the background image conversion step is calculated from the luminance value of the extracted image after the conversion by the extraction image conversion step. A process of subtracting the luminance value of the corresponding pixel in the image and calculating the observation data based on the background excluded image generated in the background excluded image generating step may be used in the observation data calculating step.

A background excluded image conversion step that converts the background excluded image into an image when a certain range is observed from above facing the liquid surface. In the observation data calculation step, the background excluded image is converted into a background excluded image after the conversion by the background excluded image converting step. A method of calculating observation data based on this may be used.

For each image obtained in the image capturing step, a method including a pre-processing step of calculating an amount representing contrast of light and dark in the image and selecting an image satisfying a predetermined condition with respect to the amount representing contrast. May be.

In the pre-processing step, the number of edges in the image is calculated as an amount representing contrast, a plurality of images satisfying a condition that the number of edges is equal to or greater than a predetermined threshold is selected, and based on the selected plurality of images A method of generating an image that is a target for extracting a region corresponding to a certain range may be used.

Further, the glass melting furnace operating method according to the present invention has an effect of deriving the degree of influence of the operating parameters of the glass melting furnace on the observation data calculated in the observation data calculation step in the above-mentioned monitoring method in the glass melting furnace. And a melting furnace control step of changing an operation parameter in which the absolute value of the degree of influence on the observation data is equal to or greater than a predetermined value when the observation data satisfies a predetermined condition It is characterized by that.

Further, the monitoring system in the glass melting furnace according to the present invention is an image photographing means for photographing an image including a reference pattern provided in the glass melting furnace and a certain range in the liquid surface of the glass raw material melted in the glass melting furnace. And an image calibration unit that extracts a region corresponding to a certain range from the captured image according to the attitude of the image capturing unit calculated using the positional deviation of the reference pattern captured in the image, and a constant Based on a plurality of extracted images extracted from a plurality of images as a region corresponding to the range, a background image creating means for creating a background image as a background of a batch mountain that is a glass raw material accumulated in a glass melting furnace, and photographing By subtracting the luminance value of the corresponding pixel in the background image from the luminance value of the pixel of the extracted image extracted as a region corresponding to a certain range from the obtained image for each pixel A difference calculating means for generating a background excluded image excluding the background from the extracted image in a state where the batch mountain and the background are reflected, and an observation data calculating means for calculating observation data regarding the batch mountain based on the background excluded image It is characterized by.

The background image creating means counts the number of pixels corresponding to each luminance value for each corresponding pixel or corresponding area of the plurality of extracted images, and based on the count result of the pixels corresponding to each luminance value, the background is generated. The background image may be created by determining the luminance value to be represented.

The difference calculation means performs, for each pixel, a process of subtracting the luminance value of the corresponding pixel in the background image from the luminance value of the pixel of the extracted image extracted as an area corresponding to a certain range from the photographed image. The configuration may be such that the background excluded image is generated by binarizing the subtraction result.

The image calibration means converts the background image into an image when a certain range is observed from above facing the liquid surface, and extracts the extracted image extracted as a region corresponding to the certain range so that the certain range faces the liquid surface. The image is converted into an image observed from above, and the difference calculation means subtracts the luminance value of the corresponding pixel in the background image after conversion by the image calibration means from the luminance value of the extracted image after conversion by the image calibration means. The observation data calculating unit may calculate the observation data based on the background excluded image generated by the difference calculating unit.

The image calibration unit converts the background excluded image generated by the difference calculation unit into an image when a certain range is observed from above facing the liquid surface, and the observation data calculation unit converts the background after the conversion by the image calibration unit. The configuration may be such that observation data is calculated based on the excluded image.

For each image obtained by the image photographing means, an amount representing the contrast of light and dark in the image is calculated, and a preprocessing means for selecting an image satisfying a predetermined condition regarding the amount representing the contrast is provided. May be.

The preprocessing means calculates the number of edges in the image as an amount representing contrast, selects a plurality of images satisfying a condition that the number of edges is equal to or greater than a predetermined threshold, and based on the selected plurality of images A configuration may be used in which an image that is a target for extracting a region corresponding to a certain range is generated.

It may be configured to include observation data analysis means for deriving the degree of influence of the operating parameters of the glass melting furnace on the observation data calculated by the observation data calculation means.

When the observation data satisfies a predetermined condition, it may be configured to include a melting furnace control means for changing an operation parameter in which the absolute value of the degree of influence on the observation data is equal to or greater than a predetermined value. .

Moreover, the method for producing a glass article according to the present invention includes a glass melting step for producing molten glass in a glass melting furnace, a clarification step for removing bubbles of the molten glass in a clarification tank,
The method includes a molding step for molding the molten glass from which bubbles have been removed, and a slow cooling step for gradually cooling the molded molten glass. The image photographing means includes a reference pattern provided in the glass melting furnace, and glass melting. According to the image photographing step for photographing an image including a certain range on the liquid surface of the glass raw material melted in the furnace, and the posture of the image photographing means calculated using the deviation of the position of the reference pattern photographed in the image Then, an area extraction step for extracting an area corresponding to a certain range from the photographed image and a plurality of extracted images extracted from a plurality of images as an area corresponding to the certain range are loaded into the glass melting furnace. A background image creating step for creating a background image as a background of a batch mountain, which is a raw glass material, and an extracted image extracted as an area corresponding to a certain range from the photographed image A background-excluded image is generated by excluding the background from the extracted image in a state in which the batch mountain and the background are reflected by performing, for each pixel, the process of subtracting the luminance value of the corresponding pixel in the background image from the luminance value of the pixel The method includes an excluded image generating step and an observation data calculating step of calculating observation data regarding the batch mountain based on the background excluded image.

According to the glass melting furnace monitoring method and the glass melting furnace monitoring system of the present invention, it is possible to continue observation of a certain area in the glass melting furnace and to monitor the state of the batch mountain in the certain area well. Moreover, according to the manufacturing method of a glass article, a glass article can be manufactured while realizing such a good monitoring state.

In addition, according to the glass melting furnace operating method of the present invention, it is possible to clarify which operating parameter of the glass melting furnace should be adjusted according to the state of the monitored batch mountain.

The top view which shows the example of the glass melting furnace to which the monitoring system in the glass melting furnace of this invention is applied. The block diagram which shows the structural example of the monitoring system in the glass melting furnace of the 1st Embodiment of this invention. Explanatory drawing which shows the example of the picked-up image by the camera _11a . Explanatory drawing which shows the example of the matching using the example of the image of a reference pattern, and a reference pattern. 9 is a flowchart showing an example of posture estimation operation performed by posture identification means 14. Among the captured image by the camera 11 _a, schematic diagram extracted range corresponding to the liquid level of dissolved material. Explanatory drawing which shows the example of the conversion result converted so that _a viewpoint may be changed just above the fixed area | region 9a. The flowchart which shows the example of the process progress of the attitude | position determination process of a camera. The flowchart which shows the example of processing progress until observation data derivation. The flowchart which shows the example of processing progress of background image creation processing (Step S11). The histogram obtained as a result of step S24. The histogram obtained as a result of step S24. Explanatory drawing which shows the example of the image after the conversion of step S13. Explanatory drawing which shows the example of the background image after conversion of step S12. Explanatory drawing which shows the example of the image of the result of performing the process of step S14. Explanatory drawing which shows the example of the image after a binarization process. Explanatory view showing a bisected regions and the central side region of the constant region 9 _a, 9 _b of the side wall 6 side region and the glass melting furnace. The block diagram which shows the structural example of the monitoring system in a glass melting furnace in the modification of 1st Embodiment. The flowchart which shows the example of the process progress until observation data derivation | leading-out in the modification of 1st Embodiment. The block diagram which shows the structural example of the monitoring system in the glass melting furnace of the 2nd Embodiment of this invention. The graph which shows the example of the result of having calculated the influence degree of the driving parameter with respect to one observation data. The graph which shows the result of having calculated the influence degree of observation data A and B and temperature A to D with respect to one quality data. The graph which shows the change of the situation where the correlation of observation data and quality data is lost or appears newly. The schematic diagram which shows an example of the manufacturing line of the glass article used with the manufacturing method of the glass article of 3rd Embodiment. The flowchart which shows the example of the manufacturing method of the glass article of 3rd Embodiment.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
First, an example of a glass melting furnace to which the glass melting furnace monitoring system of the present invention is applied will be described. FIG. 1 is a plan view showing an example of such a glass melting furnace. The glass melting furnace 1 melts a glass raw material by heat in a space surrounded by a bottom surface, an upstream wall (upstream wall) 7, a side wall 6, a downstream wall (downstream wall) 8, and a ceiling (not shown). Let The upstream wall 7 is provided with inlets 3 _a and 3 _b for introducing raw materials, and the downstream wall 8 is provided with an outlet 4 for discharging the melted glass raw material. The side walls 6 are provided with an observation window 2 and a burner 5, respectively. Although FIG. 1 shows the case where the inlets 3 _a and 3 _b are provided, the number of inlets is not limited to two.

From the inlets 3 _a and 3 _b , a solid glass material is introduced. Since the inside of the glass melting furnace is heated by the flame blown from the burner 5, this raw material is gradually melted, and the melted raw material is gradually moved downstream and discharged from the discharge port 4. The batch pile 10 is a raw material accumulated in the glass melting furnace 1 in a solid state. The batch mountain 10 dissolves while moving downstream as time passes.

The monitoring system in the glass melting furnace of the present invention includes

cameras

11 _a and 11 _b , and monitors

certain areas

9 _a and 9 _b of the liquid level in the glass melting furnace. In Figure 1, by two constant region 9 _a, 9 _b, as the region between the side walls in the front direction of each camera of the liquid level in the furnace is covered, define two constant region 9 _a, 9 _b The case is shown as an example. The camera 11 _a captures _a certain area 9 _a on the right side when viewed from the upstream side (hereinafter simply referred to as the certain area 9 _a ), and the camera 11 _b captures the certain area 9 _b on the left side when viewed from the upstream side (hereinafter referred to as “constant area 9 _a” ) , simply referred to as a constant region _{9 b.)} to shoot. In the present invention, the case where the glass melting furnace monitoring system includes two

cameras

11 _a and 11 _b will be described as an example, but the number of cameras included in the glass melting furnace monitoring system is not limited to two.

The fixed

regions

9 _a and 9 _b are determined away from the vicinity of the inlets 3 _a and 3 _b . When shooting the immediate area of the inlets 3 _a and 3 _b as a fixed area, all the parts corresponding to the fixed area in the shot image become batch mountains, and there is a high possibility that bubbles as a background do not appear. This is because data related to batch mountains cannot be calculated.

[Embodiment 1]
FIG. 2 is a block diagram illustrating a configuration example of the glass melting furnace monitoring system according to the first embodiment of the present invention. The glass melting furnace monitoring system according to the first embodiment includes a camera 11 _a , a camera 11 _b, and an image processing device 13. The monitoring system in the glass melting furnace performs the same processing on the images taken by the

cameras

11 _a and 11 _b . Therefore, hereinafter, described with respect to the camera 11 _a, description of the camera 11 _b is omitted.

The camera 11 _a through the observation window 2 of the glass melting furnace (see FIG. 1), repeatedly photographed images of certain areas 9 _a of the liquid surface. This image is a still image. Similarly, the camera 11 _b also through the observation window 2 of the glass melting furnace (see FIG. 1), repeatedly taking a still image of a certain area 9 _b of the liquid surface. Imaging interval of the camera ₁₁ a, 11 _b may be determined in advance.

Note that the camera 11 _a shooting range (range of field of view), as well as certain areas 9 _a, the liquid surface or in the vicinity constant region 9 _a, the side wall is also housed facing the camera 11 _a. Therefore, the camera 11 _a captured image of the liquid surface and the constant region 9 _a and the vicinity thereof, are also captured the opposing sidewalls. The same applies to the camera _11b .

Images taken by the

cameras

11 _a and 11 _b are input to the image processing device 13.

The image processing apparatus 13 performs image processing on the image captured by the camera 11 _a, calculates various data relating to the batch mountain in certain areas 9 _a (e.g., data relating to the arrangement and movement). Similarly, the image processing apparatus 13 performs image processing on the image captured by the camera 11 _b, calculates various data relating to the batch mountain in certain areas 9 _b. The batch mountain data calculated based on the images taken by the

cameras

11 _a and 11 _b is hereinafter referred to as observation data.

The image processing device 13 includes a preprocessing unit 19, an image storage unit 12, a posture specifying unit 14, a background image creation unit 15, an image calibration unit 16, a difference calculation unit 17, and an observation data calculation unit 18. Prepare.

Preprocessing means 19 based on the image by the camera 11 _a is taken to generate an image of the state in which the raw material powder and a frame (flame blown out from the burner 5) Implied. When the raw powder and frame floating in the glass melting furnace are reflected in the image, the image of the batch mountain becomes unclear. Preprocessing means 19, by using a plurality of images by the camera 11 _a is taken, a batch mountain generates an image in a state in which reflected in the sharp without being affected by disturbances of the raw material powder and a frame or the like. Preprocessing means 19, the same processing is performed with respect to the image by the camera 11 _b is taken. Generating an image from which the influence of the raw material powder and the frame is removed in this way is referred to as preprocessing. Further, an image generated by the preprocessing unit 19 from a plurality of images taken by the camera may be referred to as a preprocessed image hereinafter. However, the preprocessed image is the same as the image captured by each camera, except that the effect of the raw material powder and the frame is removed to make the batch mountain clearer. It may be noted. That is, it may be referred to as a photographed image in the same way as the image itself photographed by the camera. Preprocessing means 19, preprocessing image obtained on the basis of the camera 11 _a, and, the pre-processing image obtained on the basis of the camera 11 _b, respectively, are stored in the image storage unit 12.

Depending on the glass melting furnace, pre-treatment may not be necessary at all, or part of the pre-treatment may not be necessary. For example, in a glass melting furnace in which the influence of the frame is small or the raw material powder that floats is small, the pretreatment may not be performed. In that case, the image processing apparatus 13 may store the images input from the

cameras

11 _a and 11 _b in the image storage unit 12 as they are.

The image storage unit 12 is a storage device that stores an image. As described above, when the preprocessing means 19 performs the preprocessing on the images input from the

cameras

11 _a and 11 _b , the preprocessed image obtained by the preprocessing is stored. Further, when the preprocessing is not performed, the images input from the

cameras

11 _a and 11 _b are stored as they are.

Hereinafter, the case where the preprocessing unit 19 performs preprocessing and the image storage unit 12 stores the preprocessed image will be described as an example.

The posture identifying unit 14, from the image captured by the camera 11 _a (preprocessed image in this example) to identify the orientation of the camera 11 _a. Here, the posture means the position and orientation of the camera. The posture identifying unit 14 performs the same processing with respect to the camera 11 _b.

3, (in this example, the preprocessed image by the camera 11 _a is generated based on the captured image) captured image by the camera 11 _a is an explanatory diagram showing an example of. The captured image is an image taken of a certain region 9 _a direction. The image captured by the camera 11 _a, the other part above the liquid surface 25 in a batch mountain 10, is reflected also part of the opposing side walls 6 and an observation window 2. The images on the side wall 6 and the observation window 2 are used to specify the orientation and position of the camera (camera posture). That is, the boundary lines (grooves) between the bricks forming the side walls 6, the intersections between the boundary lines, and the corners (corner parts) of the observation window 2 appear as characteristic patterns in the captured image. Hereinafter, such a characteristic pattern is referred to as a reference pattern. The reference pattern needs to be a pattern in which a similar pattern does not exist in the same image when taken. For example, if a combination of the shape of a corner such as a window, a line or a point is a characteristic pattern, such a combination may be used as a reference pattern. Further, as will be described later, the posture specifying unit 14 may sequentially update the image stored as the image of the reference pattern. If the posture of the camera does not change, the reference pattern appears at a substantially constant position (coordinates) in the captured image. On the other hand, when the posture of the camera changes during cleaning or the like, the position of the reference pattern in the captured image also changes. The posture identifying unit 14 based on the position of the reference pattern in the image captured by the camera 11 _a, determines the presence or absence of displacement of the posture of the camera 11 _a. In other words, the reference pattern is used to determine whether or not a camera position shift has occurred. Note that the coordinates representing the position in the image are hereinafter referred to as image coordinates.

Also, it is preferable that a plurality of reference patterns exist in the image from the viewpoint of increasing the reliability of the determination of the camera posture deviation.

The posture specifying means 14 stores the image of the reference pattern and the image coordinates of the reference pattern in the captured image. The image coordinates of the reference pattern may be image coordinates of the center position of the reference pattern, for example. The posture identifying section 14 may, for example, an image of the point 21 _a and around the corner of the viewing window 2 stores an image of the reference pattern, and stores the image coordinates of the position. An example of an image of a reference pattern in this case and an example of matching using the reference pattern are shown in FIG. FIG. 4A shows an example of a reference pattern image. FIG. 4B shows an example of a captured image that is matched with the reference pattern. FIG. 4B illustrates a captured image similar to that in FIG. 4B, the same elements as those shown in FIG. 3 are denoted by the same reference numerals, and the description thereof is omitted. In FIG. 4A, the reference pattern is shown larger than the captured image for easy understanding. The posture specifying unit 14 performs pattern matching between the captured image and the stored image of each reference pattern, and specifies the image coordinates of the portion in the captured image corresponding to each stored reference pattern image. To do. The posture identifying section 14 determines its image coordinates, by comparing the image coordinates are stored, whether the deviation occurs in the posture of the camera 11 _a. In the pattern matching, the similarity that is a similar index value is calculated.

For example, the posture specifying unit 14 performs pattern matching between the image of the reference pattern illustrated in FIG. 4A and the captured image illustrated in FIG. b) reference) is specified, and the image coordinates of the portion 81 (for example, the center coordinates of the portion 81 in the captured image) are specified. Then, the posture identifying section 14, and the coordinates, by comparing the image coordinates stored in advance, it may be determined whether the deviation occurs in the posture of the camera 11 _a.

A characteristic point used for camera posture estimation is referred to as a reference point. In the reference point group, a point in the reference pattern (for example, the point 21 _{a at} the corner of the observation window 2) may be included. FIG. 3 illustrates the case where the points 21 _a to 21 _e are used as reference points. The posture specifying means 14 stores the image coordinates of the reference point and the three-dimensional coordinates of the reference point in real space as information on the reference point. Since the posture specifying means 14 stores “the image of the reference pattern and its image coordinates” and “the image coordinate and the three-dimensional coordinates of the reference point”, the relative positional relationship between the reference pattern and the reference point on the image is stored. I can judge.

The camera 11 _a is capturing images including the reference pattern and a constant region 9 _a, the processing by the camera 11 _b captures an image including the reference pattern and constant regions 9 _b corresponds to an image capturing step.

FIG. 5 is a flowchart illustrating an example of the posture estimation operation performed by the posture identification unit 14. The posture identifying section 14, and the image coordinates of the reference patterns in the captured image, as described above, by comparing the image coordinates are stored, if a shift in position of the camera 11 _a is determined to have occurred, their image coordinates Is used to calculate the amount of posture deviation (step S51). That is, the posture specifying means 14 calculates how much the reference pattern has shifted in the captured image.

Then, the posture specifying unit 14 reflects the amount of deviation of the reference pattern in the captured image on the stored image coordinates of the reference point (step S52). That is, the posture identifying unit 14, by the amount of shift image coordinates of the reference pattern in the captured image by a deviation occurs in the posture of the camera 11 _a, the image coordinates of shifting the image coordinates of each reference point (reference point Change the value of).

Then, the posture identifying unit 14 uses the image coordinates of the reference point, and a 3-dimensional coordinates of the reference point in the real space, performs camera calibration process, estimating the pose of the camera 11 _a. Specifically, the posture identifying section 14, the image coordinates of the individual reference points in various orientations of the cameras 11 _a, is calculated from the three-dimensional coordinates of each reference point in the real space (step S53). Then, the posture specifying unit 14 is configured such that the image coordinates calculated from the three-dimensional coordinates of each reference point are closest to the image coordinates of the reference point shifted in accordance with the shift of the image coordinates of the reference pattern as described above. the attitude, determines that the posture of the camera 11 _a (step S54).

Here, the camera 11 _a has been described as an example, the posture identifying unit 14, the determination and pose estimation of the presence or absence of a deviation of the orientation related to the camera 11 _b similarly performed.

Image calibration means 16, according to the posture of the camera 11 _a posture specifying unit 14 has identified, (in this example, before the processed image) in the captured image to identify the range corresponding to the constant region 9 _a in. 6, among the image taken by the camera 11 _a, a schematic diagram obtained by extracting the range corresponding to the liquid surface 25 of the dissolved material. In addition, the right side and the left side of FIG. 6 are the upstream and downstream of the glass melting furnace, respectively. Among the image of the liquid surface 25, _a range 31 enclosed by a thick solid line, corresponds to the constant region 9 _a in the real space. The image calibration unit 16 identifies and extracts _a range 31 _a corresponding to the certain region 9 _a according to the posture of the camera 11 _a .

However, the liquid level in the glass melting furnace is assumed to be constant. A certain range of area 9 _a in the height is predetermined. That is, a certain range of area 9 _a (position) is pre-defined as a position of the region in the plane of constant height in the real space. Therefore, the posture of the camera 11 _a is identified, it is possible to determine also the range corresponding to the constant region 9 _a in the image captured by the camera 11 _a. That is, the image calibration unit 16 may specify the range 31 _a in the image when the constant region 9 _a at _a constant height in the real space is projected onto the captured image of the camera 11 _a whose posture is known. .

In addition, when the liquid level in the glass melting furnace is assumed to be constant, the pixel resolution in the photographed image is investigated by investigating how many millimeters the deviation of one pixel in the photographed image is displaced in the real space. (Mm / pixel) can be grasped.

The image calibration means 16, the range 31 _a corresponding to the constant region 9 _a in the image to another process for specifying an image of the range 31 _a, in other words immediately above (the constant region 9 _a, liquid A process of converting the image into an image when observed from the upper side facing the surface is also performed. That is, the image illustrated in FIG. 6 is a picture in the case of observing a certain area 9 _a camera 11 _a viewpoint (inclination direction to the liquid surface), with respect to the scope 31 _a in the image, a certain viewpoint into image in the case of changing directly above the area 9 _a. An example of the conversion result is illustrated in FIG. Thus, the image the image calibration unit 16, with respect to the range 31 _a corresponding to the constant region 9 _a in the image, performs viewpoint conversion processing for changing the view point directly over the constant region 9 _a, which is observed from the viewpoint Should be generated.

Note that the image calibration means 16 is not limited to the range 31 _a extracted from the image captured by the camera 11 _{a that} is to be converted into an image when the fixed region 9 _a is observed from directly above. For example, the image calibration unit 16 performs the same conversion on an image obtained by image processing (for example, background image creation processing described later).

Image calibration means 16 (in this example, the preprocessed image) captured image by the camera 11 _b the same processing is performed with respect.

Background image creating means 15, using the image before the processing means range 31 extracted by the image calibration unit 16 from a plurality of pre-processing images sequentially generated by 19 a _(range 31 corresponding to a certain area 9 _a _a) Then, an image of the liquid level is created when there is no batch mountain (background image creation processing). This range 31 _a is a picture corresponding to the constant region 9 _a, the pictures in the batch mountain background bubbles. Further, since the moving speed and the dissolution rate of the batch mountain is moderate, in the range 31 _a, always (or high in frequency) is reflected batch mountain. Therefore, it is difficult to directly capture an image in which only bubbles (background) are shown as the range 31 _a corresponding to the certain region 9 _a . Therefore, the background image creating unit 15 uses the range 31 _a extracted from the plurality of images, to create a background image batch mountain does not exist.

Bubbles are present on the liquid surface where no batch pile exists. In addition, the batch mountain dissolves without gradually moving in the downstream direction. Therefore, even pixels corresponding to the batch mountain in the range 31 _a extracted from an image, would represent a range 31 _a in bubbles extracted from another image. The background image creating means 15 extracts, for each set of corresponding pixels in the range 31 _a corresponding to the fixed area 9 _a extracted from a plurality of images (in other words, pixels corresponding to the same position in the fixed area 9 _a ). By specifying the brightness corresponding to the bubbles, an image representing only the background of the batch mountain is created without the presence of the batch mountain. Here, the case where processing is performed for each set of corresponding pixels in the range 31 _a extracted from a plurality of images is taken as an example, but for each corresponding area in the range 31 _a extracted from a plurality of images, a bubble is generated. The luminance corresponding to may be specified. An area is an area formed by gathering consecutive pixels.

Background image creating means 15 (in this example, the preprocessed image) captured image by the camera 11 _b the same processing is performed with respect.

The difference calculation means 17 calculates the difference between corresponding pixels between two images. Specifically, the luminance value of the corresponding pixel in the background image is subtracted from the luminance value of each pixel of the image showing the batch mountain. By this subtraction process, an image obtained by removing the background portion from the image showing the batch mountain is obtained. However, there are some changes in the brightness of the bubbles. Therefore, the result of subtracting the luminance value of the corresponding pixel in the background image from the luminance of the pixel corresponding to the bubble in the image showing the batch mountain is not always zero. Therefore, the difference calculation means 17 subtracts the luminance value of the corresponding pixel in the background image from the luminance value of each pixel of the image showing the batch mountain, and then subtracts the subtraction result for each pixel to “0” or “1”. It is preferable to perform a process for converting the value. In this binarization process, the difference calculation means 17 rounds up the subtraction result to “1” for each pixel if the subtraction result is equal to or greater than a predetermined value, and if the subtraction result is less than the predetermined value, It may be rounded down to “0”. By performing this binarization processing, the region corresponding to the batch mountain (the region having the luminance value “1”) and the region corresponding to the background (the region having the luminance value “0”) are more clearly distinguished. can do.

The observation data calculation means 18 calculates the observation data of the batch mountain from the image from which the background portion is removed and the portion corresponding to the batch mountain is left. Examples of observation data include, for example, the position of the tip of the batch crest, the movement speed of the batch crest, the dissolution rate of the batch crest (batch crest reduction rate), the occupancy ratio of the batch crest in each of the

constant regions

9 _a and 9 _b , and the like. It is done. Further, regarding these observation data, a difference between _a value in the fixed region 9a and _a value in the fixed region _9b may be calculated, and the difference may be used as the observation data.

Further, the constant region 9a is divided into two parts, _a side wall region and a central region in the width direction of the glass melting furnace, and the ratio of batch occupancy ratios in the two regions (hereinafter referred to as an internal / external ratio). .) May be calculated as observation data. Likewise, for certain regions 9 _b, and side wall regions, bisected into an inner region of the glass melting furnace, the ratio of occupancy of the batch mountain in the two regions (inside and outside ratio) as the observation data You may calculate.

The pre-processing means 19, the posture specifying means 14, the background image creation means 15, the image calibration means 16, the difference calculation means 17 and the observation data calculation means 18 are realized by a CPU of a computer that operates according to a program, for example. In this case, for example, the CPU reads a program stored in a program storage device (not shown) of the computer, and the CPU follows the program in accordance with the preprocessing means 19, the posture specifying means 14, the background image creating means 15, and the image calibration means 16. The difference calculating unit 17 and the observation data calculating unit 18 may be operated.

Next, the operation will be described.
First, preprocessing by the preprocessing unit 19 will be described. The camera 11 _a periodically photographs the direction of the fixed area 9 _a and inputs the images sequentially to the preprocessing means 19. Preprocessing means 19, a constant cycle (e.g., the period of a few seconds) for each, based on the plurality of images input from the camera 11 _a in its cycle, generating a preprocessed image. Specifically, the preprocessing unit 19 counts the number of edges in the image for each image input within one period. Note that an edge is a line that appears in an image. For example, the region to be counted for the number of edges in the image may be limited to _a region corresponding to the wall surface and a region corresponding to the certain region 9a. The processing cycle by pre-processing means 19 short, each image input from the camera 11 _a within that period, many cases the large number of is reflected batch mountain does not change. Also, the fact that the number of batch peaks in the image does not change means that the number of edges should be maintained at a certain level if there is no influence of the frame or the raw material powder. By utilizing this, the pre-processing means 19, from among a plurality of images input from the camera 11 _a in one cycle, a plurality of continuous images is maintained the number of count result is often a state of the edge select. For example, a predetermined threshold value may be used as a reference for determining the number of edge counts in the image. Specifically, the preprocessing unit 19 determines that the number of edges in the image is large when the number of edges obtained as a count result satisfies a condition that the number of edges is equal to or greater than a predetermined threshold. Then, an image having the number of edges equal to or greater than a threshold value may be selected. Further, when the number of edges obtained as a count result is less than the threshold, the preprocessing unit 19 determines that the number of edges in the image is small and does not select an image having the number of edges less than the threshold. Alternatively, the judgment criterion for the number of edges may be changed according to the count result of the number of edges in each input image.

In the above description, the case where the preprocessing unit 19 selects a plurality of continuous images has been described as an example. However, the plurality of images selected by the preprocessing unit 19 may not be continuous images.

Further, the preprocessing means 19 may calculate an amount representing the contrast between light and dark in the image and select an image satisfying a predetermined condition regarding the amount representing the contrast. The number of edges described above is an example of an amount representing the contrast between light and dark in an image. The condition that the number of edges is equal to or greater than the threshold is an example of a condition that is predetermined with respect to an amount that represents contrast between light and dark. An example in which the preprocessing means 19 selects an image by a method other than the image selection method based on the number of edges is shown below. For example, the pre-processing means 19, a camera 11 for each input image from _a, a quantity representing the contrast of the brightness of the image may be calculated standard deviation of the luminance values. At this time, the preprocessing means 19 may calculate the standard deviation of the luminance value of each pixel included in the entire image. Alternatively, in the image, an area where the boundary line between the bricks is captured may be determined in advance, and the preprocessing unit 19 may calculate the standard deviation of the luminance value in that area in the image. In addition, as an example of the condition for selecting an image, an image from the occurrence of an event in which the amount representing the contrast of the image is lower than the amount representing the contrast of the previous image by a certain value or more until the lapse of the certain time is excluded. And a condition of selecting an image remaining without being excluded. For example, when this condition is adopted and the standard deviation of the luminance value is calculated as an amount representing the contrast between light and dark, the preprocessing unit 19 uses the standard deviation of the luminance value of the previous image as the standard deviation of the luminance value in a certain image. In the case of lowering by a certain value or more, an image generated until a certain period elapses from that point is excluded from subsequent processing targets, and an image remaining without being excluded may be selected. Then, the preprocessing unit 19 generates a preprocessed image from the selected plurality of images. It should be noted that the fact that the amount representing the contrast between light and dark in the image has decreased by a certain value or more means that the contrast has suddenly decreased, and it can be considered that a phenomenon such as the raw material powder rising has occurred.

In the following description, a case where the preprocessing unit 19 selects an image based on the number of edges in the image will be described as an example.

The preprocessing unit 19 generates a preprocessed image by determining the luminance value of each pixel in the preprocessed image using the selected plurality of images. In a plurality of selected images, attention is paid to corresponding pixels (pixels having the same image coordinates), and the minimum luminance value among the pixels is specified. Then, the preprocessing unit 19 determines the luminance value as the luminance value of the corresponding pixel in the preprocessed image. For example, the pre-processing unit 19 reads the luminance values of the image coordinates of each image selected (x _{1, y} _1), specifies the minimum value of the luminance values in the image coordinate (x _{1, y} _1). Then, the preprocessing unit 19 determines the minimum luminance value as the luminance value at the image coordinates (x ₁ , y ₁ ) of the preprocessed image. The preprocessing unit 19 performs this processing for each pixel. The preprocessing unit 19 stores the generated preprocessed image in the image storage unit 12. The preprocessing means 19 repeats this processing at a constant cycle. Thus, the preprocessed image by the camera 11 _a is generated based on the captured images are sequentially accumulates the image storage unit 12.

Incidentally, in the preprocessing, of the plurality of images input from the camera 11 _a, the "plurality of continuous images are kept count result many state of the edge" other images may be ignored.

Here, the camera 11 _a is explained an exemplary case of using the image taken, the camera 11 _b also periodically capturing a predetermined region 9 _b direction, and inputs the image sequence, the pre-processing means 19 . The preprocessing unit 19 similarly generates a preprocessed image from the image captured by the camera 11 _b and stores it in the image storage unit 12.

It can be said that a plurality of continuous images maintaining a state in which the edge count result is large are images in which the frame and the raw material powder are not shown so much. This is because, in an image in which many frames and floating raw material powder are shown, batch peaks and side walls become unclear and the number of edges in the image decreases. In addition, when a frame is captured, the luminance value of a portion corresponding to the frame in the image is a high value. Accordingly, as described above, by selecting a plurality of images in which the frame and the raw material powder are not shown so much, and by specifying the minimum luminance value among the corresponding pixels in those images, the frame and the raw material powder are not shown. The luminance value in the state image can be selected. Since determining the preprocessed image as an image having such a brightness value, a part of the image the camera 11 _a is taken, even captured the raw material powder and a frame suspended in the furnace, such raw material powder and the frame Can be generated. That is, it is possible to obtain an image that clearly shows the batch mountain to be monitored. The operation in which the preprocessing means 19 generates a preprocessed image corresponds to a preprocessing step.

In addition, as already explained, it is not necessary to perform the pretreatment as described above in a glass melting furnace in which the influence of the frame is small or the raw material powder that floats is small. In that case, the image processing device 13 may store the images taken by the

cameras

11 _a and 11 _b in the image storage unit 12 as they are.

Next, an operation in which the posture specifying unit 14 determines the posture of the camera will be described. Here, as an example the case of determining the orientation of the camera 11 _a, the attitude determination processing of the camera 11 _b is similar. FIG. 8 is a flowchart illustrating an example of processing progress of the camera posture determination processing. In this example, the case where the posture specifying unit 14 stores images of a plurality of reference patterns and their image coordinates will be described as an example.

As described above, the preprocessing unit 19, every fixed period (e.g., period of a few seconds), generates a preprocessed image from the image by the camera 11 _a is taken, and stores the image in the image storage unit 12.

The posture identifying means 14 (in this example, the front camera 11 _a is generated based on the captured image processed image) a plurality of captured images stored in the image storage unit 12 reads, the posture of the camera 11 _a A process for determining whether or not a deviation has occurred is periodically performed. However, the processing cycle of the posture specifying unit 14 is longer than the processing cycle of the preprocessing unit 19 compared to the processing cycle of the preprocessing unit 19 being, for example, several seconds. For example, the processing cycle of the posture specifying means 14 may be several hours.

The posture identifying unit 14 determines that becomes the processing start timing, reads the captured image of the latest predetermined number stored in the image storage unit 12 (front camera 11 _a is generated based on the captured image processed image) . The predetermined number may be determined in advance. The posture specifying means 14 generates an average image of the read predetermined number of photographed images (preprocessed images) (step S1). Specifically, the posture specifying unit 14 calculates an average value of luminance values for each corresponding pixel with respect to a predetermined number of read captured images, generates an image using the average value as a luminance value, An average image may be used. In this example, the case where an average image is generated is illustrated, but an intermediate value of luminance values may be calculated for each corresponding pixel, and an image (intermediate value image) having the intermediate value as a luminance value may be generated.

Further, in this example, the case where an average image is generated from a plurality of images in step S1 is exemplified, but the processing after step S2 may be performed on one image stored in the image storage unit 12. That is, the process of step S1 may be omitted.

The posture specifying unit 14 performs pattern matching on the plurality of reference patterns stored in advance by the posture specifying unit 14 on the average image generated in step S1 (step S2). In this example, the case where the calculated similarity value becomes smaller as the degree of similarity between images is higher will be described as an example. In step S <b> 2, the posture specifying unit 14 calculates the similarity between the reference pattern image stored in advance and each unit in the average image. Then, the position in the image having the highest degree of similarity (in this example, the similarity is the smallest value) is specified. For example, if the image of the reference pattern illustrated in FIG. 4 and the image coordinates thereof are stored in advance, the posture specifying unit 14 has a similarity value with the image of the reference pattern illustrated in FIG. 4 from the average image. A location that is equal to or less than the threshold is specified, and further, a location having the smallest similarity value is specified from those locations. This portion is a portion corresponding to the reference pattern in the average image. Furthermore, the posture specifying unit 14 specifies, for example, the image coordinates of the central pixel at the specified location. That is, the posture specifying unit 14 specifies a portion most similar to the image of the reference pattern illustrated in FIG. 4 from the average image, for example, specifies the image coordinates of the central pixel. The posture specifying means 14 performs this process for each image of the reference pattern stored in advance.

Calculating the degree of similarity may be performed by a known method. For example, examples of the similarity include SSD (Sum? Of? Squared? Difference?) And SAD (Sum? Of? Absolute? Difference). The SSD is a total value of the squares of differences in luminance values between corresponding pixels in a pair of images to be subjected to similarity calculation. Therefore, the posture specifying unit 14 calculates the SSD by calculating the square of the difference between the luminance values for each pair of corresponding pixels in the pair of images as the similarity calculation target, and further calculating the total value thereof. do it. Further, SAD is a total value of absolute values of differences in luminance values between corresponding pixels in a pair of images to be subjected to similarity calculation. Accordingly, the posture specifying unit 14 calculates the absolute value of the difference in luminance value for each pair of corresponding pixels in the pair of images that are the similarity calculation target, and further calculates the total value thereof, thereby calculating the SAD. What is necessary is just to calculate. When the image that is the similarity calculation target is a binary image, the posture specifying unit 14 performs XOR (eXclusive OR) for each pair of corresponding pixels in the pair of images that are the similarity calculation target. Or the total value thereof, and the calculation result may be used as the similarity. The total value of SSD, SAD, and XOR for each pair of pixels is a similarity that decreases as the degree of similarity between images increases.

Note that, in this example, a case is described in which the degree of similarity in which the value decreases as the degree of similarity between images increases, but other degrees of similarity may be used. For example, the posture specifying means 14 may calculate a normalized cross-correlation (NCC) as the similarity. The normalized cross-correlation has a value closer to 1 as the degree of similarity between images increases. Therefore, when the normalized cross-correlation is calculated as the similarity, the posture specifying unit 14 may specify a location where the similarity (normalized cross-correlation) value is closest to 1.

Next, the posture specifying unit 14 stores in advance, for each reference pattern, the image coordinates of the portion having the highest degree of similarity (the portion having the smallest similarity value in this example) specified in step S2. the difference between the image coordinates of which was the reference pattern (i.e., distance) is calculated, based on the distance, determines whether the deviation occurs in the posture of the camera 11 _a (step S3). The posture specifying unit 14 compares the distance between the image coordinates specified in step S2 and the feature coordinates stored in advance with a threshold value. If the distance between the coordinates is equal to or larger than the threshold value, the posture of the camera is shifted. If it determines with having arisen and the distance between coordinates is less than a threshold value, what is necessary is just to determine that there is no shift | offset | difference in the attitude | position of a camera. Since the posture specifying unit 14 stores a plurality of reference patterns in advance, the distance between the coordinates (the difference between the image coordinates specified in step S2 and the feature coordinates stored in advance) for each reference pattern. calculate. A criterion for comparing the plurality of distances with the threshold value and determining whether or not the posture of the camera has shifted is not particularly limited. For example, it may be determined that a deviation has occurred in the posture of the camera on condition that a predetermined number or more of a plurality of inter-coordinate distances obtained by calculation for each reference pattern is equal to or greater than a threshold value. Alternatively, it may be determined that a deviation has occurred in the posture of the camera on condition that all the inter-coordinate distances are equal to or greater than the threshold value. Here, two criteria are illustrated, but it may be determined whether or not a deviation has occurred in the posture of the camera according to other criteria.

When it is determined that the camera posture is deviated (Yes in step S3), the posture specifying unit 14 specifies the image coordinates of the reference pattern image and the image coordinate stored in advance in step S2. By substituting the image coordinates, the image coordinates in the set of the reference pattern image and the image coordinates stored are updated (step S4). That is, the posture specifying unit 14 sets the image coordinates of the part specified as the part corresponding to the reference pattern in the average image (in the above example, the image coordinates of the central pixel of the part) in combination with the image of the reference pattern. The stored image coordinates are updated as the image coordinates. By the process of step S4, the coordinates (image coordinates) of the reference pattern in the average image are updated in accordance with the deviation of the camera posture. However, the posture specifying means 14 is also used in the process of step S5 for the image coordinates before update. The image coordinates before update are also stored until used in step S5.

After step S4, the posture specifying means 14 estimates the posture of the camera _11a using the reference point (step S5). In step S5, the posture specifying means 14 may perform the following processing. Based on the image coordinates of the reference pattern before update (the image coordinates of the reference pattern stored in advance) and the image coordinates of the updated reference pattern, the posture specifying unit 14 determines how much the reference pattern is in the image, Calculate in which direction it has shifted. In the case where there are a plurality of reference patterns, for example, the average deviation amount for each reference pattern or the average deviation direction may be calculated, and the average value may be used as the deviation amount and deviation direction of the reference pattern. Alternatively, the reference pattern deviation amount and the deviation direction may be determined based on other standards. The posture specifying means 14 shifts the image coordinates of the reference points stored in advance according to the direction and amount of deviation of the reference pattern. That is, the coordinate value of the image coordinates of the reference point is updated in accordance with the deviation of the reference pattern before and after the update. Then, the posture identifying section 14, the image coordinates of the individual reference points in various orientations of the cameras 11 _a, is calculated from the three-dimensional coordinates of each reference point in the real space. Then, the posture identifying section 14, the image coordinates calculated from the three-dimensional coordinates of each reference point, identifies the closest become posture image coordinates of each reference point of the updated, its posture is a posture of the camera 11 _a Is determined. Then, the designated estimation process (that is, the process of step S5) is terminated.

When it is determined that there is no deviation in the camera posture (No in step S3), the posture specifying unit 14 stores in advance an image of a portion corresponding to the reference pattern in the average image specified in step S2. The image of the reference pattern that has been updated is updated (step S6). That is, in step S2, an image of a portion specified as a portion corresponding to the reference pattern in the average image is extracted, and the image is stored as a new reference pattern image. The posture specifying means 14 performs this process for each reference pattern. By the processing in step S6, the image of the reference pattern and the image of the reference pattern in the set of image coordinates stored in advance by the posture specifying means 14 are updated.

The state of the side wall in the glass melting furnace gradually changes, and the degree of similarity between the position corresponding to the reference pattern in the image and the image of the reference pattern stored in the posture specifying means 14 may decrease. For example, as shown in FIG. 4, even when the pattern near the corner of the observation window is used as the reference pattern, the raw material powder gradually adheres to the corner portion, so that an image of the reference pattern portion in the photographed image is obtained. May gradually change from a right-angled corner image to a rounded corner image. Assuming that the stored reference pattern image is not updated, this change becomes large, and when a new image is taken, the image cannot be matched with the reference pattern image illustrated in FIG. . However, in step S5, the next pattern matching can be accurately performed by updating the stored reference pattern image based on the pattern matching result in the average image. For example, the reference pattern image illustrated in FIG. 4 stored in advance can be gradually updated to a reference pattern image with rounded corners. As a result, the next pattern matching can be accurately performed, and the posture determination of the camera can also be accurately performed.

The posture specifying means 14 may perform the processes after step S1 at regular intervals on the preprocessed image generated based on the image taken by the camera _11a and stored in the image storage means 12. Similarly, with respect to the preprocessed image generated based on the image photographed by the camera _11b and stored in the image storage unit 12, the processing after step S1 may be performed at regular intervals.

Even if the pre-processing is not performed and the images captured by the

cameras

11 _a and 11 _b are stored in the image storage unit 12 as they are, the posture specifying unit 14 performs the processing on the images captured by the camera 11 _a. Thus, the processing after step S1 may be performed at regular intervals. Then, likewise, the image of the camera 11 _b is taken, every predetermined period, it may be performed the processing at and after Step S1.

Next, an operation for creating an arrangement image (see FIG. 7) of batch mountains in a certain region when observed from directly above and calculating observation data will be described. FIG. 9 is a flowchart showing an example of processing progress of this operation. Here, (in this example, the camera 11 _a is preprocessed image which is generated based on the captured image) captured image by the camera 11 _a is will be described as an example when the image processing apparatus 13 performs processing for the image processing device 13 (in this example, the preprocessed image generated based on the image by the camera 11 _b is taken) image captured by the camera 11 _b the same processing is performed with respect to.

First, the image calibration unit 16 (in this example, the preprocessed image) captured image by the camera 11 _a stored in the image storage unit 12 reads a plurality sequentially from newer. The number of captured images to be read at this time may be determined in advance. Then, the image calibration unit 16 extracts from the respective captured images, a range 31 corresponding to a certain area _{9 a} in the real space _a (see FIG. 6) (step S10). Extracted range 31 _a an image shown (hereinafter, referred to as an extracted image.) Is a batch mountain image to background foam. Here for convenience is described a case where the posture of the camera 11 _a is not changed as an example, when the posture of the camera 11 _a is changed, the image calibration unit 16, when the image capturing of the camera 11 _a of Based on the posture, a range 31 _a corresponding to the certain region 9 _a in the real space may be extracted from the captured image.

Note that when the preprocessing is not performed and the images captured by the

cameras

11 _a and 11 _b are respectively stored in the image storage unit 12 as they are, the image calibration unit 16 captures the image by the camera 11 _a in step S10. It is only necessary to read a plurality of photographed images stored in the image storage unit 12 in order from the newest one and extract an extracted image from each photographed image. Further, with regard photographed image by the camera 11 _b is stored as it is in the image storage unit 12 imaging the same. The other points are the same whether or not pre-processing is performed. Step S10 corresponds to a region extraction step.

Next, the background image creating means 15 creates an image when no batch mountain exists based on the extracted images respectively extracted from the plurality of photographed images. That is, a background image as a background of the batch mountain is created (step S11). In step S11, a background image having pixels having the same image coordinates as that of the extracted image extracted from the latest photographed image and a luminance value of the pixel representing a bubble is created. Step S11 corresponds to a background image creation step.

FIG. 10 is a flowchart showing an example of processing progress of the background image creation processing in step S11.

In the background image creation process, the background image creation means 15 selects individual pixels in the extracted image extracted from the latest photographed image, and the brightness of the selected pixel and the pixels in the other extracted images corresponding to the selected pixel. Based on the value, a luminance value representing the background in the selected pixel is determined. As a result, a background image is obtained when no batch mountain exists. Hereinafter, this process will be described with reference to FIG. Note that, here, a case where the luminance value representing the background is determined for each pixel will be described as an example, but the background image creating unit 15 determines the luminance value representing the background for each individual area in the extracted image. May be.

The background image creating means 15 selects one pixel from the pixels of the extracted image extracted from the latest photographed image (step S21). Then, the background image creating unit 15, from the extracted image extracted from other photographed image in step S10 (see FIG. 9), the pixel corresponding to the selected pixel (i.e., the same position in a predetermined region 9 _a (Corresponding pixel) is extracted (step S22).

Next, the background image creating means 15 targets the pixel selected in step S21 and the pixel in the other extracted image corresponding to the pixel (that is, the pixel obtained in step S22) for each luminance value. The number of pixels corresponding to the luminance value is counted (step S24). It can be said that the process of step S24 is a histogram creation process.

Subsequently, the background image creating means 15 evaluates the variation of the luminance value within the luminance value range in which the pixel count number (frequency) is increased (step S25). The range of luminance values in which the number of counts is increased is, for example, a range in which luminance values in which the number of counts is equal to or greater than a threshold value (threshold value determined for the count number) continue. 11 and 12 are histograms obtained as a result of step S24. In the example shown in FIG. 11, the range of the luminance value in which the pixel count is large is k ₁ to k ₂ . In the example shown in FIG. 12, the range of luminance values in which the number of pixel counts is large is k ₃ to k ₄ . As the evaluation value for evaluating the variation, for example, the standard deviation or variance of the luminance values of the pixels counted within such a range may be used. Alternatively, the width of the luminance value range in which the pixel count number is increased may be used as the evaluation value. In step S25, such an evaluation value may be calculated. In the case of calculating the standard deviation, variance, or the width of the range of luminance values in which the number of pixel counts are increased as the evaluation value, the smaller the evaluation value, the smaller the variation in the luminance value. Another index value may be used as an evaluation value of variation.

After step S25, the background image creating means 15 determines whether or not there is a large variation in luminance value within the range of luminance values in which the number of pixel counts is large, based on the evaluation value calculated in step S25. (Step S26). In step S26, it is only necessary to determine whether or not the variation is large by comparing a predetermined threshold value (threshold value for the evaluation value of variation) with the evaluation value. For example, when the standard deviation of luminance values is calculated as an evaluation value, if the evaluation value is equal to or greater than a threshold value (threshold value determined for the evaluation value), it is determined that the variation is large, and if the evaluation value is less than the threshold value It can be determined that the variation is small. The threshold value may be determined in advance according to an index value (standard deviation, variance, etc.) adopted as the evaluation value.

When it is determined that the variation in luminance value is small (No in step S26), the background image creating unit 15 determines the most frequent luminance value within the range of luminance values where the count value is large (step S28). FIG. 11 is an example of a histogram when the variation in luminance value is small. Taking FIG. 11 as an example, the range of luminance values where the count value is large is k ₁ to k ₂ , and the most frequent luminance value within this range (the luminance value where the pixel count number is maximum). ) Is S. Therefore, the background image creation means 15 specifies the value of S in step S28. Then, the value of S is determined as the luminance value in the pixel at the coordinate selected in step S21. If the selected coordinate has a small variation in luminance value, it can be said that the batch mountain is not captured at that coordinate and the background continues to be captured. Therefore, when the variation is small, the mode luminance value S can be determined as the luminance value of the bubble as the background as described above. Note that instead of the mode luminance value S as described above, the average value of the luminance values of the pixels corresponding to the luminance value range k ₁ to k ₂ in which the count value is large is calculated, and the average value is You may determine as a luminance value showing a background. Alternatively, the median value of the luminance value range k ₁ to k ₂ may be determined as the luminance value representing the background.

On the other hand, when it is determined that the variation in the luminance value is large (Yes in step S26), the background image creation unit 15 corresponds to a luminance value larger than the discrimination reference value within the range of the luminance value where the count value is large. An average value of luminance values of each pixel is calculated (step S27). FIG. 12 is an example of a histogram when the variation in luminance value is large. Taking FIG. 12 as an example, the range of luminance values where the count value is large is k ₃ to k ₄ . Further, it is assumed that the discrimination reference value is T. At this time, the background image creating means 15 calculates the average value of the luminance values of the pixels corresponding to the range up to k ₄ where the luminance value is greater than T. Then, the background image creating means 15 determines the average value as the luminance value at the pixel at the coordinate selected in step S21. If the selected coordinates have a large variation in luminance value, it can be said that a batch mountain appears in the coordinates or a background bubble appears. And the brightness value of a bubble is larger than the brightness value of a batch mountain. Therefore, the average of the luminance values of the pixels corresponding to the range larger than the discrimination reference value as described above can be determined as the luminance value of the bubbles serving as the background. Instead of calculating the average value as described above, in a range larger than determination reference value within a range of luminance values that count value becomes a number (in the range of T ~ k ₄ in the example shown in FIG. 12) The mode luminance value may be determined, and the mode luminance value may be determined as the luminance value in the pixel at the selected coordinate. Alternatively, the median value in the range from T to k ₄ may be determined as the luminance value in the pixel at the selected coordinate.

Note that the discriminant reference value is a threshold value for separating a large variation range (luminance value range in this example) into two, and is a threshold value in the discriminant analysis binarization method described in Non-Patent Document 3. Applicable. Therefore, the threshold value that maximizes the variance ratio between the intra-class variance and the inter-class variance for the background area and the batch mountain area may be used as the discrimination reference value T.

Here, the case where the luminance value range k ₃ to k ₄ is divided into two classes by the discriminant analysis binarization method is shown, but the luminance value range k ₃ to k ₄ is divided into two classes by other methods. You may divide into. For example, the luminance value range k ₃ to k ₄ may be divided into two classes by a mode method, a method of fitting two normal distributions, or the like. Then, from the class with the higher luminance value, the luminance value in the pixel at the selected coordinate may be determined in the same manner as described above.

The background image creating unit 15 performs the above-described processing described with reference to the flowchart of FIG. 10 for each pixel, and uses the luminance value obtained in step S27 or step S28 for the background image corresponding to the pixel selected in step S21. It is determined as the luminance value of the pixel. As a result, in the extracted image extracted from the latest photographed image, an image from which the batch mountain is removed is obtained. The image is a background image when viewed from the perspective of the camera 11 _a.

Further, the background image creating means 15 may determine a luminance value representing the background for each area obtained by dividing the extracted image. In this case, in step S21, the background image creating means 15 selects one area from the extracted image extracted from the latest photographed image. There is no particular limitation on how to define the area. Then, the background image creating unit 15, in step S22, from the extracted image extracted from another captured image, and extracts an area corresponding to the selected area (area corresponding to the same portion of the constant region 9 _a) . In step S24 and subsequent steps, a histogram is created for each pixel belonging to the area selected in step S21 and the area corresponding to the area (the area obtained in step S22), and the evaluation value of the variation in luminance value is calculated. And the luminance value may be calculated according to whether or not the variation is large (steps S24 to S28). The background image creating means 15 performs this processing for each area obtained by dividing the extracted image, and uses the luminance value obtained in step S27 or step S28 for the background image corresponding to the area selected in step S21. What is necessary is just to determine as a luminance value of each pixel in an area.

After the background image creation process, the process proceeds to step S12 (see FIG. 9). In step S12, the image calibration unit 16, a background image obtained by the background image creating process (step S11), and converts the image when observed from directly above the constant region 9 _a (step S12). That is, for the background image obtained in step S11, performs viewpoint conversion processing for changing the viewpoint from the position of the camera 11 _a directly above the constant region 9 _a, creating a background image when viewed from the viewpoint. As a result, with no batch mountain exists in a certain area 9 _a, an image of when observed from directly above the constant region 9 _a is obtained. Step S12 corresponds to a background image conversion step.

Next, the image calibration unit 16, in step S10, the extraction image extracted from the latest captured image is converted into an image when observed from directly above the constant region 9 _a (step S13). That is, for extraction image extracted from the latest captured image, performs viewpoint conversion processing for changing the viewpoint from the position of the camera 11 _a directly above the constant region 9 _a, converts the image when viewed from the viewpoint. This converted image shows a batch mountain and background. The conversion process in steps S12 and S13 is a similar conversion process. Step S13 corresponds to an extracted image conversion step.

In addition, when the sizes of the images after the conversion in steps S12 and S13 are different, the image calibration unit 16 may perform correction so that the sizes of the images after the conversion in steps S12 and S13 are made uniform.

The processing after step S14 described later may be executed using the converted image obtained in steps S12 and S13 as it is.

Alternatively, just above each for detecting a latest photographed image, the image processing apparatus 13 executes the processing from step S10 to step S13, the image calibration unit 16, an image (constant region 9 _a obtained in step S12 a background image) when observed from the respective images (images when observed from directly above the constant region 9 _a) obtained in step S13, may be plural storage. Then, the image calibration means 16 selects the latest predetermined number of images obtained each time step S12 is executed, synthesizes the selected images (for example, generates an average image), and similarly performs step S13. It is also possible to select the latest predetermined number of images obtained each time execution is performed and synthesize the selected images. Then, the combined image of the images obtained each time executing the step S12 (the background image when observed from directly above the constant region 9 _a), the synthesis of the image obtained in each execution of step S13 the image (constant region 9 The processing after step S14, which will be described later, may be executed using an image obtained by observing _a from directly above.

Most of the raw materials in the solid state exist below the liquid level. Therefore, using the images obtained one by one in steps S12 and S13, the composite image of a plurality of images obtained in each step S12 and the image obtained in each step S13 are obtained rather than performing the processing in the next step S14 and subsequent steps. In addition, it is easier to grasp the whole image of the solid-state raw material from the obtained observation data by performing the processing after the next step S14 using the composite image of the plurality of images. Therefore, as described above, a plurality of images obtained each time Step S12 is executed are combined, and similarly, a plurality of images obtained each time Step S13 is executed are combined, and these combined images are used. It is preferable to execute the processing after step S14.

When combining a plurality of images obtained each time step S13 is executed, the image calibration unit 16 calculates, for example, the average value of the luminance values of the corresponding pixels in the plurality of images, and uses the average value as the combined image. The luminance value of the corresponding pixel in can be used. What is necessary is just to produce | generate a synthesized image by performing this process for every pixel and determining each luminance value of a synthesized image. Further, instead of the average value of the luminance values of the corresponding pixels, a minimum value may be specified for the luminance value of each corresponding pixel, and the minimum value of the luminance value may be used as the luminance value of the corresponding pixel in the composite image.

The image calibration unit 16 may generate a composite image by performing the same process when combining a plurality of images obtained each time step S12 is executed.

In addition, when calculating the movement speed of the batch mountain as the observation data, the processing after step S14 is performed using the images obtained in steps S12 and S13 without generating the above-described composite image. Just do it. Further, when calculating the movement speed of the batch mountain, the processes after step S10 are performed using the images themselves taken by the

cameras

11a and 11b.

Next, the difference calculation means 17 calculates a difference in luminance value between corresponding pixels between the image after the conversion at Step S13 and the background image after the conversion at Step S12 (Step S14). Here, the image after the conversion in step S13 may be one image obtained in step S13 or a composite image of a plurality of images obtained each time step S13 is executed. Similarly, the background image after the conversion in step S12 may be one image obtained in step S12 or a composite image of a plurality of images obtained each time step S12 is executed. .

In step S14, the difference calculation means 17 subtracts the luminance value of the pixel of the background image after the conversion in step S12 from the luminance value of the pixel of the image after the conversion in step S13 (an image showing the batch mountain and the background). . The difference calculation means 17 performs this subtraction process for each pair of corresponding pixels.

FIG. 13 shows an example of the image after the conversion in step S13. In this image, the background and the batch mountain 10 are shown. FIG. 14 shows an example of the background image after the conversion in step S12. FIG. 15 shows an example of an image obtained as a result of performing the process of step S14 on these two images. As described above, since there is some change in the brightness of the bubble, the brightness value of the pixel corresponding to the background is not always 0 after the process of step S14.

After step S14, the difference calculation means 17 performs a binarization process on the image (see FIG. 15) obtained in step S14 (step S15). That is, for each pixel in the image, the difference calculation means 17 performs a process of replacing a luminance value equal to or higher than a predetermined threshold for binarization processing with “1” and replacing a luminance value less than the threshold with “0”. . Since the luminance value of the pixel corresponding to the background has become a value near 0 by the subtraction process in step S14, it becomes “0” by the binarization process. Further, the luminance value of the pixel corresponding to the batch crest 10 is set to “1” by the binarization process because the value is not greatly reduced by the subtraction process in step S14. As a result, the luminance value of the pixel corresponding to the background is “0”, and the luminance value of the pixel corresponding to the batch mountain 10 is “1”. An example of an image after binarization is shown in FIG. Binarization after image represents the position of the batch mountain in certain areas 9 _a which is created based on the extracted image extracted from the latest captured image. Incidentally, this image shows a state observed from directly above the point of view of a certain region 9 _a, information of the height of the batch mountain does not include. The difference calculation means 17 stores the image generated in step S15 (hereinafter, binarized image). Steps S14 and S15 correspond to a background excluded image generation step.

After step S15, the observation data calculating unit 18, using the binary image generated in step S15, and calculates the observation data of the batch pile present in the constant region 9 _a (step S16). However, in step S16, the observation data may be calculated using not only the most recently generated binarized image but also a binarized image that continues from the past. Further, here, it has been described for the generation of the binarized image related to a certain area 9 _a, the image processing apparatus 13, based on the image captured by the camera 11 _b, also generates binarized image related to a certain region 9 _b. The observation data calculation means 18 may calculate observation data based on the binarized images of the fixed

areas

9 _a and 9 _b . Step S16 corresponds to an observation data calculation step.

Hereinafter, an example of the observation data calculated in step S16 is shown. As an example of the observation data, the inside / outside ratio of each of the

constant regions

9 _a and 9 _b can be given. FIG. 17 is an explanatory diagram showing a region in which the

constant regions

9 _a and 9 _b are divided into two equal parts: a region on the side wall 6 side and a region on the center side of the glass melting furnace. Elements similar to those shown in FIG. 1 are given the same reference numerals as those in FIG.

Regions

51 and 52 is an area in which bisects constant region 9 _a to the region of the side wall 6 side region and the central side, the region 51 is the region of the side wall 6 side, region 52 in the center side region is there. Similarly, regions 41 and 42 is an area in which bisects constant region 9 _b in the region and the central side region of the side wall 6 side, region 41 is the region of the side wall 6 side, the region 42 is the center side It is an area. Observation data calculating means 18, as the inner and outer ratio for constant region 9 _a, the occupancy rate of the batch mountain region 51 may be calculated evaluation value representing the ratio of occupancy of batch mountain region 52. Similarly, as the inner and outer ratio for constant region 9 _b, may be calculated and occupancy batch mountain region 41, an evaluation value representing the ratio of the occupancy of the batch mountain region 42.

For example, when the occupancy rate of the batch mountain in the side wall 6 side region (that is, the region 51 or the region 41) is Q, and the batch mountain occupancy rate in the central region (ie, the region 52 or the region 42) is R. The observation data calculation means 18 may calculate the evaluation value represented by the following formula (1) as the internal / external ratio. However, Q and R are expressed as percentages and are values in the range of 0 to 100, respectively.

Internal / external ratio = (RQ) / (R + Q + α) Equation (1)

In the formula (1), α is a constant, and for example, α may be 100. In this case, the inside / outside ratio is a value in the range of −0.5 to 0.5. The observation data calculation means 18 may calculate the inside / outside ratio for each of the fixed

regions

9 _a and 9 _b .

If the raw material in the solid state is too close to the side wall 6 side, the raw material may flow out of the glass melting furnace without being melted, and in this case, the quality of the glass is deteriorated. It is possible to confirm whether or not the solid state raw material is too close to the side wall 6 by the inside / outside ratio. When it is determined that the raw material in the solid state is too close to the side wall 6 side, the glass melting furnace may be operated so that the batch mountain approaches the center.

Moreover, the observation data calculation means 18 may calculate the occupancy ratio of the batch mountain in each of the

constant regions

9 _a and 9 _b .

Further, the observation data calculation means 18 may calculate the tip position of the batch mountain (for example, the coordinates of the tip position of the batch mountain) in each of the fixed

regions

9 _a and 9 _b .

In addition, there is a case where the tip position of the batch mountain is not shown in the binarized image due to the state of the batch mountain or the spread of the frame. In this case, the observation data calculating means 18 divides in a direction perpendicular to the traveling direction of the raw material obtained by dissolving a predetermined region 9 _a, calculates an area of the batch mountain in each divided region. Then, assuming that the change in the area of the batch mountain from the upstream divided region toward the downstream divided region is a linear change, the position where the batch mountain area becomes 0 is calculated, and the position is the tip of the batch mountain. The position may be determined. The same applies to the constant region _9b .

If the tip of the batch mountain extends too far downstream, it may flow out undissolved. When the tip position of the batch peak calculated by the observation data calculation means 18 is excessively extended to the downstream side, the glass melting furnace may be operated so that the tip position of the batch peak returns to the upstream side.

Further, the observation data calculating means 18, the value of the observational data in the right-hand constant region 9 _a as viewed from the upstream side, the difference between the value of the observational data when viewed from the upstream side in the predetermined region 9 _b of the left observation data May be calculated as For example, the observation data calculating means 18, the occupancy rate of the batch mountain in certain areas 9 _a, may calculate the difference between the occupancy of the batch mountain in certain areas 9 _b. Further, the observation data calculating means 18, the tip position of the batch mountain in certain areas 9 _a, may calculate the difference between the tip position of the batch mountain in certain areas 9 _b. Hereinafter, the difference in the values of the observation data in the fixed

regions

9 _a and 9 _b is referred to as a left / right difference. By calculating this left / right difference as one of the observation data, it is possible to confirm whether the state of the solid raw material is not biased between the right side and the left side as viewed from the upstream side. For example, it is possible to confirm that the melting progresses only on one of the right side and the left side when viewed from the upstream side, and that the melting is delayed on the other side, and the glass melting furnace is operated according to the situation. Can be judged.

For example, when it is determined that the dissolution of the raw material in either one of the

constant regions

9 _a and 9 _b is delayed due to the difference between the left and right calculated in step S < _b > 16, it is determined that the raw material is delayed in the constant region. An operation such as increasing the amount of fuel input to the burner on the near side wall (ie, increasing the burner's heating power) may be performed.

In the above example, the case where the left-right difference regarding the occupancy ratio and the tip position of the batch mountain is calculated has been described. However, the observation data calculation unit 18 may calculate the left-right difference regarding other observation data.

Further, the occupancy ratio of the batch mountain, the tip position, and the left-right difference between them may be calculated from the most recent binary image, but may be calculated from the composite image of the most recent binary images. Also good. In addition, as already explained, from the viewpoint of grasping the whole image of the raw material in the solid state, the images obtained for each step S12 are synthesized, and the images obtained for each step S13 are synthesized, and those images are obtained. It is preferable to generate a binarized image by performing the processing after step S14 using the composite image.

Further, the observation data calculation means 18 may calculate the movement speed of the entire batch mountain based on the position of the same batch mountain in a plurality of continuous binarized images and the photographing interval of the camera. Since the movement of the entire batch mountain is slow, there is little change in the position of the same batch mountain in a plurality of continuous binarized images. Therefore, the observation data calculation unit 18 may determine that the batch mountains having the closest position coordinates are the same batch mountain in a plurality of continuous binarized images. Then, the movement distance of the batch mountain may be calculated from the change in the coordinates of the same batch mountain, and the movement speed of the entire batch mountain may be calculated from the movement distance and the photographing interval. In this example, the movement speed of one batch mountain is regarded as the movement speed of the entire batch mountain. In addition, when calculating the speed of a batch mountain as observation data, the process after step S10 is performed using the image itself which the

camera

11a, 11b image | photographed. Furthermore, the process after step S14 is performed using one image obtained in step S12 and one image obtained in step S13.

Further, the observation data calculation means 18 may calculate the movement direction of the batch mountain based on the same batch mountain position in a plurality of continuous binarized images.

Further, the observation data calculation means 18 may calculate the batch mountain reduction rate from a plurality of continuous binarized images. For example, the observation data calculation unit 18 may determine the same batch mountain in each continuous binarized image, and calculate the reduction rate of the area and length of the batch mountain in each binarized image. In calculating the length reduction rate, the reduction rate may be calculated based on the length along the flow direction of the raw material, or based on the length along the direction perpendicular to the flow direction of the raw material. Thus, the decrease rate may be calculated.

The observation data calculation means 18 preferably uses a plurality of continuous binarized images when calculating the movement speed of the entire batch mountain, the movement direction of the batch mountain, the reduction rate of the batch mountain, and the like. A plurality of binarized images may be used.

This batch mountain reduction rate is considered to have a correlation with the batch mountain height reduction rate, and the batch mountain height can be determined from the batch mountain reduction rate. If the batch crest is too high, it takes time to dissolve and the tip position is extended.

Further, the observation data calculation means 18 may calculate the direction of each batch mountain (the direction in which the batch mountain extends) from the binarized image. Such a direction may be expressed in advance by determining a reference direction and an angle formed with the reference direction.

Further, the observation data calculation means 18 may calculate the size of each batch mountain from the binarized image.

Further, the observation data calculation means 18 may calculate an evaluation value for evaluating the gas blowing state in the batch mountain based on the binarized image and the image obtained in step S13. When gas is blown out from the batch mountain, a hole depressed on the surface of the batch mountain is observed in the image and looks rough. Therefore, the observation data calculation means 18 determines a region corresponding to the batch mountain from the images obtained in step S13 using the binarized image, calculates a standard deviation of luminance values in the region, and calculates the standard. The deviation may be an evaluation value of the gas blowing state.

In addition, the portion depressed by the gas blowout is observed as a black region. Therefore, the observation data calculation means 18 determines a region corresponding to the batch mountain from the images obtained in step S13 using the binarized image, counts the total number of black pixels in the region, and counts the count. The result may be an evaluation value of the gas blowing state.

In Non-Patent Document 1 and Patent Document 1, it is described that the occupancy ratio of the batch mountain and the tip position (the most downstream position) of the batch mountain are evaluated. By measuring various observation data such as difference, batch crest speed and moving direction, batch crest reduction rate, individual batch crest direction and size, evaluation value of gas blowing state in batch crest, etc. The mountain can be quantitatively evaluated stably. Moreover, based on the result, high-quality glass can be produced by appropriately operating the glass melting furnace.

Further, according to the present invention, the posture specifying unit 14 performs pattern matching of the reference pattern on the captured image (more specifically, the average image of the captured image), and sets the image coordinates of the reference pattern in the captured image. Based on this, it is determined whether or not there is a shift in the posture of the camera. If it is determined that a shift has occurred, the posture (position and orientation) of the camera is specified using the amount of shift in the posture. Then, the image calibration unit 16, based on the attitude of the camera, extracting a range corresponding to the constant region 9 _a, 9 _b in real space from a captured image. Further, the background image creating unit 15 creates a background image from the extracted image, and the image calibration unit 16 performs viewpoint conversion processing for changing the viewpoint of the extracted image and the background image from the position of the camera to a position directly above a certain region. Then, the difference calculation means 17 calculates the difference between the two luminance values. Therefore, even if the posture of the camera changes during cleaning or the like, observation of a certain region in the glass melting furnace can be continued well.

Also, the preprocessing means 19 selects a plurality of consecutive images that maintain a state where the edge count result is large from among a plurality of images input from the camera as preprocessing. Then, the preprocessing unit 19 pays attention to the corresponding pixel in the selected plurality of images, specifies the minimum luminance value among the pixels, and determines the luminance value of the corresponding pixel in the preprocessed image. Determine as The preprocessing means 19 performs this processing for each corresponding pixel. Depending on the image taken by the camera, raw powder floating in the furnace may appear, the frame may appear, and the background and batch mountains may be blurred, but by performing the pretreatment as described above, It is possible to create an image that is less affected by disturbance such as a frame or raw material powder. Then, using such an image, the process after step S10 (see FIG. 9) can be performed to obtain a good background image that is not affected by disturbance or an image that shows only batch mountains. It is possible to accurately monitor the condition of the batch mountain in a certain area.

Therefore, according to this embodiment, it is possible to continue observation of a certain region in the glass melting furnace and to monitor the state of the batch mountain in the certain region well. In addition, as already demonstrated, when monitoring the glass melting furnace with little influence of raw material powder or a flame | frame, it is not necessary to perform a pre-processing. In that case, the processing after step S10 (see FIG. 9) may be performed using the image itself generated by photographing the inside of the furnace by the camera.

Further, in the present embodiment, the posture specifying unit 14 performs pattern matching of a plurality of reference patterns on the captured image, and specifies the posture of the camera. Thus, since a plurality of reference patterns are used, the reliability of the determination of the posture deviation of the camera is increased.

Next, a modification of the first embodiment will be described. In the first embodiment described above, the conversion process (steps S12 and S13, see FIG. 9) is performed on the extracted image extracted from the background image and the latest photographed image, and the difference is calculated (step). S14. See FIG. 9). The conversion process may be performed after the process of calculating the difference between the pixels is performed first. FIG. 18 is a block diagram showing a configuration example of the monitoring system in the glass melting furnace in such a modification of the first embodiment. Each means shown in FIG. 18 is the same as each means shown in FIG. 2, and is denoted by the same reference numerals as in FIG. However, in this modification, the flow of various images is partially different from that in the first embodiment, and therefore the arrows indicating the flow of various images are different from those in FIG. FIG. 19 is a flowchart illustrating an example of a processing progress until observation data calculation in the modification of the first embodiment. The same processes as those described in the first embodiment are denoted by the same reference numerals as those in FIG.

In this modified example, after steps S10 and S11, the difference calculation means 17 performs a luminance value between corresponding pixels between the extracted image extracted from the latest photographed image and the background image created in step S11. Is calculated (step S31). At this time, the difference calculating means 17 subtracts the luminance value of the pixel of the background image from the luminance value of the pixel of the extracted image (image showing the batch mountain and the background) extracted from the latest photographed image. The difference calculation means 17 performs this subtraction process for each pair of corresponding pixels. As a result, it is possible to obtain an image of a certain region viewed from the viewpoint of the camera and having the background removed. However, in the above subtraction result, the luminance value of the pixel corresponding to the background is not always 0.

Therefore, the difference calculation means 17 performs binarization processing on the image obtained in step S31 after step S31 (step S32). As a result, the binary image is an image of a certain region viewed from the viewpoint of the camera, the luminance value of the pixel corresponding to the background is “0”, and the luminance value of the pixel corresponding to the batch mountain 10 is “1”. A converted image is obtained. Steps S31 and S32 correspond to a background excluded image generation step.

After step S32, the image calibration unit 16 performs viewpoint conversion processing for changing the viewpoint from the position of the camera to a position directly above a certain region with respect to the binarized image generated in step S32 (step S33). As a result, a binarized image similar to the binarized image obtained in step S15 (see FIG. 9) already described is obtained. Step S33 corresponds to a background excluded image conversion step.

After step S32, the observation data calculation means 18 calculates the observation data of the batch mountain existing in a certain region using the binarized image after the conversion process in step S32 (step S16). This process is the same as the process of step S16 already described.

[Embodiment 2]
FIG. 20 is a block diagram illustrating a configuration example of the glass melting furnace monitoring system according to the second embodiment of the present invention. The same components as those in the first embodiment are denoted by the same reference numerals as those in FIG. Glass melting furnace monitoring system of the second embodiment includes a camera 11 _a, and the camera 11 _b, and an image processing unit 13 _a. The image processing apparatus 13 _a includes observation data analysis in addition to the preprocessing means 19, the image storage means 12, the posture specifying means 14, the background image creation means 15, the image calibration means 16, the difference calculation means 17, and the observation data calculation means 18. Means 61 and melting furnace control means 62 are provided. The image processing apparatus 13 _a may be configured by adding the observed image processing apparatus for a glass melting furnace monitoring system data analysis means 61 and the melting furnace control device 62 shown in FIG. 18.

The observation data analysis means 61 determines the degree of correlation between various observation data whose values are calculated by the observation data calculation means 18 and various operating parameters of the glass melting furnace. In other words, the observation data analysis means 61 derives the degree of influence of various operating parameters of the glass melting furnace on various observation data whose values are calculated by the observation data calculation means 18. As examples of observation data, the occupancy ratio of the batch mountain in each of the

constant regions

9 _a and 9 _b , the tip position of the batch mountain, and the left-right difference between the observation data, the internal / external ratio in the

constant regions

9 _a and 9 _b , the batch mountain The moving speed, the reduction rate of the batch mountain, and the like can be mentioned, but the observation data is not limited to these. The operation parameters include burner fuel combustion conditions (for example, combustion amount), raw material input conditions (for example, input amount), batch / cullet ratio, etc., but the operation parameters are not limited to these.

The observation data analysis means 61 determines the degree of correlation between the observation data and the operation parameter by, for example, principal component analysis and multivariate analysis (for example, multiple regression analysis). For example, the observation data analysis means 61 performs principal component analysis to obtain a principal component, and performs multivariate analysis using the principal component. And the observation data analysis means 61 derives the influence degree of each parameter by using the coefficient used in said process. Specifically, the influence level of the parameter is the degree of influence of the operation parameter on the observation data. The process in which the observation data analysis means 61 derives the parameter influence degree corresponds to an influence degree derivation step.

FIG. 21 is a graph showing an example of the result of calculating the influence of the operation parameter on one observation data (here, observation data A). FIG. 21 shows the correlation between the observation data A, the input parameters A (referred to as the amount of raw material input), the input conditions B, and the combustion parameters A to D, which are operation parameters. The vertical axis in FIG. 21 represents the degree of influence of each operation parameter. The combustion parameters A to D are the amount of combustion in the burner at each location. If the influence value of the operation parameter is positive, there is a positive correlation with the observation data, and if the influence value of the operation parameter is negative, there is a negative correlation with the observation data. In addition, the larger the absolute value of the influence value, the stronger the degree of correlation between the operation parameter and the observation data.

For example, from the result shown in FIG. 21, if the input condition A (raw material input amount) is increased, it means that the value of the observation data A is also increased. Further, if the combustion parameter A is increased, the value of the observation data A is decreased.

The melting furnace control means 62 refers to the observation data calculated by the observation data calculation means 18, and if the observation data has reached a value at which the operating state of the glass melting furnace should be changed, the observation data is between the observation data. Change the correlated operating parameters. Here, the operation parameter having a correlation with the observation data is, for example, an operation parameter in which the absolute value of the degree of influence on the observation data is equal to or greater than a predetermined value. For example, if the value of the observation data exceeds the upper limit value and is too high, the value of the operating parameter having a positive correlation with the observation data is decreased, or the value of the observation data is Or increase the value of the operating parameter having a negative correlation. Further, for example, when the value of the observation data is less than the lower limit value and is too low, the value of the operation parameter having a positive correlation with the observation data is increased, or the observation data and The value of the operating parameter having a negative correlation between the two is decreased. As a specific example, when it is determined that there is a negative correlation between the occupancy rate of the batch mountain as the observation data and the furnace temperature as the operation parameter, and the occupancy rate of the batch mountain exceeds the upper limit value, The furnace control means 62 may operate the glass melting furnace so as to increase the furnace temperature. That is, the burner's heating power may be increased. Thus, the process in which the melting furnace control means 62 changes the operation parameter corresponds to a melting furnace control step.

Also, the melting furnace control means 62 may output an alarm when the observation data value exceeds the upper limit value or falls below the lower limit value.

The operator may change the operating parameters of the glass melting furnace. In this case, the melting furnace control means 62 may not be provided. In this case, the operator refers to the observation data calculated by the observation data calculation means 18 and the influence degree between the observation data calculated by the observation data analysis means 61 and the operation parameters. What is necessary is just to judge how to change.

According to the present embodiment, the observation data analysis means 61 calculates the degree of influence indicating the degree of correlation of the operation parameter with respect to the observation data, so which operation parameter of the glass melting furnace is determined according to the monitored batch mountain state. It can be clarified whether adjustment is necessary.

Furthermore, by providing the melting furnace control means 62, the glass melting furnace can be automatically controlled to an appropriate state without depending on the operator.

In the above description, the case where the observation data analysis unit 61 calculates the influence of the operation parameter on the observation data is shown. In addition, when quality data (for example, the number of bubbles) indicating the quality of the raw material state is obtained, the observation data analyzing means 61 has an influence degree indicating the degree of correlation between the observation data and the operation parameters with respect to the quality data. It may be calculated. This degree of influence may also be performed by, for example, principal component analysis and multivariate analysis. In addition, it means that the state of a kiln is so bad that there are many foam numbers.

FIG. 22 is a graph showing the result of calculating the influence of observation data A and B and the operating parameters temperatures A to D on the number of bubbles as one quality data. The observation data A and B are data calculated by the observation data calculation unit 18 using a binarized image generated based on the captured image. The temperatures A to D are values obtained by measuring the temperature at each location of the glass melting furnace. Also in the example shown in FIG. 22, if the influence value is positive, there is a positive correlation between the observation data and temperature and the quality data, and if the influence value is negative, the observation data and temperature And quality data have a negative correlation. In addition, the larger the absolute value of the influence value, the greater the degree of correlation.

For example, it can be seen from the results shown in FIG. 22 that the larger the values of the observation data A and B, the greater the number of bubbles (the quality is worse). Moreover, it turns out that the number of bubbles increases, so that the value of the temperature A is low.

Even if it is determined that there is a correlation between certain observation data and quality data under a certain condition, there is a correlation between other observation data and its quality data under another condition. It may be determined that there is. FIG. 23 is a graph showing changes in the situation in which the correlation between the observation data and the quality data is lost or newly appears. The left vertical axis shown in FIG. 23 indicates the value of the observation data. The vertical axis on the right side shows the value of quality data (here, the number of bubbles). The horizontal axis represents the passage of time. In the example shown in FIG. 23, a correlation was observed between the observation data A and the quality data until the middle of the measurement period, but the correlation was lost in the latter half. Further, there was no correlation between the observation data B and the quality data until the middle of the measurement period, but a correlation between the observation data B and the quality data was recognized in the latter half.

Therefore, it is preferable that the observation data analysis means 61 repeatedly calculate the degree of influence between the observation data and the quality data.

In the second embodiment, an operation parameter that has a correlation with observation data is determined based on the degree of influence calculated by the observation data analysis unit 61, and the operation parameter is changed according to the observation data. . When the operator can determine which operation parameter to operate by referring to the binarized image, the operator may increase or decrease the operation parameter by referring to the binarized image. For example, when it is determined from the binarized image that the dissolution of the right batch mountain as viewed from the upstream wall is slow, the operator may increase the heating power of the right burner as viewed from the upstream wall.

Further, in the above embodiments, the camera 11 _a is arranged at a position to be photographed from directly above the constant region 9 _a, the camera 11 _b may be arranged in a position for imaging from above the constant region 9 _b . Also in this case, it is assumed that characteristic objects (for example, side walls, burners, etc.) are included in the shooting range, and a reference pattern and a reference point are also shot. Thus, the camera 11 _a is arranged at a position to be photographed from directly above the constant region 9 _a, when the camera 11 _b is disposed at a position to be photographed from directly above the constant region 9 _b, constant viewpoint area 9 _a need not perform the viewpoint conversion processing or changing just above the constant region 9 _b. That is, the viewpoint conversion process in steps S12 and S13 (see FIG. 9) need not be performed. Further, in the process progress shown as a modification of the embodiment (see FIG. 19), the viewpoint conversion process in step S33 may not be performed.

[Embodiment 3]
Next, a glass article manufacturing method will be described as a third embodiment of the present invention. The glass melting furnace monitoring method described in the first embodiment is applied to the glass article manufacturing method of the present invention. Furthermore, the determination of the degree of correlation between the observation data and the operation parameter described in the second embodiment and the operation parameter change process may be applied to the method for manufacturing a glass article of the present invention. FIG. 24 is a schematic diagram illustrating an example of a production line for glass articles used in the method for producing a glass article of the present embodiment. In FIG. 24, the

cameras

11 _a and 11 _b and the image processing device 13 are not shown, but the

cameras

11 _a and 11 _b are disposed in the vicinity of the glass melting furnace 1. An image processing device 13 is also arranged. However, the arrangement position of the image processing apparatus 13 is not limited. It may also be an image processing apparatus 13 _a described in the second embodiment are arranged.

A glass melting furnace 1 and a clarification tank 30 are provided in a production line for glass articles. In addition, the kind of clarification tank 30 is not limited. The clarification tank 30 may be a depressurization type clarification tank in which the inside of the tank is depressurized to remove bubbles. Alternatively, the clarification tank 30 may be a high-temperature type clarification tank in which the inside of the tank is heated to remove bubbles.

The glass melting furnace 1 (see FIG. 24 and FIG. 1) melts a glass raw material and changes it into a molten glass 71. In FIG. 24, the illustration of the batch mountain is omitted. The clarification tank 30 removes bubbles generated in the molten glass 71. The molten glass from which the bubbles have been removed moves to a forming step and a slow cooling step.

FIG. 25 is a flowchart showing an example of a method for manufacturing a glass article of the present embodiment. First, a glass raw material is charged into the glass melting furnace 1. The glass melting furnace 1 includes a burner 5 (see FIG. 1), and maintains the interior of the glass melting furnace 1 at a high temperature. And the molten glass 71 is manufactured by heating the raw material of glass in the glass melting furnace 1 (step S91, glass melting step).

At step S91, the camera 11 _a, 11 _b is taken to the glass melting furnace 1, the image processing apparatus 13 performs the same processing as in the first embodiment with respect to the resulting image. That is, processes such as steps S51 to S54 (see FIG. 5), steps S1 to S6 (see FIG. 8), steps S10 to S16 (see FIG. 9 or FIG. 19), steps S21 to S28 (see FIG. 10), and the like are performed. By this processing, observation data is obtained, and the inside of the glass melting furnace 1 can be monitored well. The second image processing apparatus 13 _a described in the embodiment of, as in the second embodiment, the degree of correlation between the observed data and the operating parameters of the glass melting furnace 1 determined, the glass melting furnace 1 The operation parameter may be changed.

The molten glass 71 manufactured in step S91 is caused to flow into the clarification tank 30. Bubbles exist in the molten glass 71, and a bubble layer (not shown) is generated on the surface of the molten glass 71. Inside the clarification tank 30, bubbles of the molten glass 71 are removed (step S92, clarification step).

After step S92, the molten glass from which bubbles have been removed is formed (step S93, forming step). In the forming step, for example, the molten glass may be formed by a float process. Specifically, the molten glass 71 from which bubbles have been removed is floated on molten tin (not shown) and is advanced in the conveying direction to form a continuous plate-like glass ribbon. At this time, in order to form a glass ribbon having a predetermined plate thickness, a rotating roll is pressed against both side portions of the glass ribbon, and the glass ribbon is stretched outward in the width direction (direction perpendicular to the conveying direction).

Next, the glass ribbon formed in step S93 is gradually cooled (step S94, slow cooling step). In the slow cooling step, the glass ribbon is pulled out from the molten tin, and the glass ribbon is gradually cooled inside a slow cooling furnace (not shown). Even after transporting to the outside of the slow cooling furnace, the glass ribbon is gradually cooled to near normal temperature.

After the slow cooling step, the glass ribbon solidified in the slow cooling step is processed as necessary (step S95, processing step). Examples of processing in step S95 include cutting and polishing. However, the present invention is not limited to cutting and polishing, and other processing may be performed.

According to the method for manufacturing a glass article of the present embodiment, the glass article can be manufactured while observing a certain region in the glass melting furnace well. In particular, the image processing apparatus 13 _a is, as in the second embodiment, the degree of correlation between the observed data and the operating parameters of the glass melting furnace 1 determined, by changing the operating parameters of the glass melting furnace 1, the furnace A glass article can be manufactured by operating the glass melting furnace 1 with an appropriate operation parameter according to the observation result.

Although this application has been described in detail and with reference to specific embodiments, it will be apparent to those skilled in the art that various changes and modifications can be made without departing from the spirit and scope of the invention.
This application is based on a Japanese patent application filed on May 6, 2011 (Japanese Patent Application No. 2011-103601), the contents of which are incorporated herein by reference.

Industrial applicability

The present invention is suitably applied to a glass melting furnace monitoring system for monitoring batch hills in a glass melting furnace.

11 _a , 11 _b camera 12 image storage means 13, 13 _a image processing device 14 posture specifying means 15 background image creation means 16 image calibration means 17 difference calculation means 18 observation data calculation means 19 preprocessing means 61 observation data analysis means 62 melting Furnace control means

Claims

An image capturing step in which the image capturing means captures an image including a reference pattern provided in the glass melting furnace and a certain range in the liquid surface of the glass raw material melted in the glass melting furnace;
A region extracting step of extracting a region corresponding to the predetermined range from the captured image according to the attitude of the image capturing means calculated using a shift in the position of the reference pattern captured in the image;
A background image creating step for creating a background image as a background of a batch mountain that is a glass raw material accumulated in a glass melting furnace based on a plurality of extracted images extracted from a plurality of images as a region corresponding to the predetermined range; ,
By performing, for each pixel, a process of subtracting the luminance value of the corresponding pixel in the background image from the luminance value of the pixel of the extracted image extracted as an area corresponding to the certain range from the photographed image, the batch mountain and A background-excluded image generating step for generating a background-excluded image in which the background is excluded from the extracted image in which the background is reflected;
An observation data calculation step of calculating observation data relating to the batch mountain based on the background excluded image.
In the background image creation step,
The number of pixels corresponding to each luminance value is counted for each corresponding pixel or corresponding area of a plurality of extracted images, and the luminance value representing the background is determined based on the count result of the pixels corresponding to each luminance value. The method for monitoring a glass melting furnace according to claim 1, wherein a background image is created.
In the background exclusion image generation step,
A process for subtracting the luminance value of the corresponding pixel in the background image from the luminance value of the pixel of the extracted image extracted as an area corresponding to a certain range from the photographed image is performed for each pixel, and the subtraction result for each pixel is binary. The method for monitoring a glass melting furnace according to claim 1 or 2, wherein a background-excluded image is generated by converting the background exclusion image.
A background image converting step for converting a background image into an image when a certain range is observed from above facing the liquid surface;
An extraction image conversion step of converting the extracted image extracted as a region corresponding to the certain range into an image when the certain range is observed from above facing the liquid surface;
In the background excluded image generation step, a process of subtracting the luminance value of the corresponding pixel in the background image after conversion by the background image conversion step from the luminance value of the extraction image after conversion by the extraction image conversion step,
The glass melting furnace monitoring according to any one of claims 1 to 3, wherein in the observation data calculation step, observation data is calculated based on the background exclusion image generated in the background exclusion image generation step. Method.
A background exclusion image conversion step for converting the background exclusion image into an image when a certain range is observed from above facing the liquid surface;
The observation method in the glass melting furnace according to any one of claims 1 to 3, wherein in the observation data calculation step, observation data is calculated based on the background excluded image converted by the background excluded image conversion step. .
A pre-processing step of calculating an amount representing contrast of light and dark in the image for each image obtained in the image capturing step and selecting an image satisfying a predetermined condition with respect to the amount representing the contrast. The monitoring method in a glass melting furnace of any one of Claims 1-5.
In the preprocessing step, the number of edges in the image is calculated as an amount representing contrast, a plurality of images satisfying a condition that the number of edges is equal to or greater than a predetermined threshold is selected, and the selected plurality of images are selected. The glass melting furnace monitoring method according to claim 6, wherein an image to be a target for extracting an area corresponding to a certain range is generated based on the glass melting furnace monitoring method.
The degree of the influence of the operating parameters of the glass melting furnace on the observation data calculated in the observation data calculation step in the monitoring method in the glass melting furnace according to any one of claims 1 to 7. A degree of influence deriving step for deriving
A melting furnace control step of changing an operating parameter in which the absolute value of the degree of the influence on the observation data is equal to or more than a predetermined value when the observation data satisfies a predetermined condition, Glass melting furnace operation method.
Image photographing means for photographing an image including a reference pattern provided in the glass melting furnace and a certain range in the liquid surface of the glass raw material melted in the glass melting furnace,
An image calibrating unit that extracts a region corresponding to the certain range from the captured image according to the attitude of the image capturing unit calculated using a shift in the position of the reference pattern captured in the image;
A background image creating means for creating a background image as a background of a batch mountain that is a glass raw material accumulated in a glass melting furnace based on a plurality of extracted images extracted from a plurality of images as a region corresponding to the predetermined range; ,
By performing, for each pixel, a process of subtracting the luminance value of the corresponding pixel in the background image from the luminance value of the pixel of the extracted image extracted as an area corresponding to the certain range from the photographed image, the batch mountain and Difference calculating means for generating a background excluded image excluding the background from the extracted image in a state in which the background is reflected;
An observation data calculation means for calculating observation data related to the batch mountain based on the background excluded image.
The background image creation means counts the number of pixels corresponding to each luminance value for each corresponding pixel or corresponding area of the plurality of extracted images, and based on the count result of the pixels corresponding to each luminance value, The glass melting furnace monitoring system according to claim 9, wherein a background image is created by determining a luminance value representing
The difference calculation means performs, for each pixel, a process for subtracting the luminance value of the corresponding pixel in the background image from the luminance value of the pixel of the extracted image extracted as a region corresponding to a certain range from the photographed image. The monitoring system in a glass melting furnace of Claim 9 or Claim 10 which produces | generates a background exclusion image by binarizing the subtraction result of.
The image calibrating means converts the background image into an image when a certain range is observed from above facing the liquid surface, and extracts the extracted image extracted as a region corresponding to the certain range as the liquid range. Converted to an image when observed from above facing the surface,
The difference calculation means performs a process of subtracting the brightness value of the corresponding pixel in the background image after conversion by the image calibration means from the brightness value of the extracted image after conversion by the image calibration means,
The glass melting furnace monitoring system according to any one of claims 9 to 11, wherein the observation data calculation unit calculates observation data based on a background excluded image generated by the difference calculation unit.
The image calibration unit converts the background excluded image generated by the difference calculation unit into an image when a certain range is observed from above facing the liquid surface,
The glass melting furnace monitoring system according to any one of claims 9 to 11, wherein the observation data calculation unit calculates observation data based on a background excluded image after conversion by the image calibration unit.
And a pre-processing unit that calculates an amount representing a contrast of light and dark in the image for each image obtained by the image photographing unit and selects an image that satisfies a predetermined condition with respect to the amount representing the contrast. The glass melting furnace monitoring system according to any one of claims 9 to 13.
The preprocessing means calculates the number of edges in the image as an amount representing contrast, selects a plurality of images satisfying a condition that the number of edges is equal to or greater than a predetermined threshold, and sets the selected plurality of images The glass melting furnace monitoring system according to claim 14, wherein an image to be a target for extracting a region corresponding to a certain range is generated based on the glass melting furnace monitoring system.
The observation data analyzing means for deriving the degree of the influence of the operating parameter of the glass melting furnace on the observation data calculated by the observation data calculating means. The monitoring system in the glass melting furnace described.
The melting furnace control means for changing an operating parameter in which the absolute value of the degree of influence on the observation data is equal to or greater than a predetermined value when the observation data satisfies a predetermined condition. The monitoring system in the glass melting furnace described.
A glass melting step for producing molten glass in a glass melting furnace;
A clarification step of removing bubbles of the molten glass in a clarification tank;
Forming step of forming molten glass from which bubbles have been removed;
And a slow cooling step of slowly cooling the molded molten glass,
An image capturing step in which the image capturing means captures an image including a reference pattern provided in the glass melting furnace and a certain range in the liquid surface of the glass raw material melted in the glass melting furnace;
A region extracting step of extracting a region corresponding to the predetermined range from the captured image according to the attitude of the image capturing means calculated using a shift in the position of the reference pattern captured in the image;
A background image creating step for creating a background image as a background of a batch mountain that is a glass raw material accumulated in a glass melting furnace based on a plurality of extracted images extracted from a plurality of images as a region corresponding to the predetermined range; ,
By performing, for each pixel, a process of subtracting the luminance value of the corresponding pixel in the background image from the luminance value of the pixel of the extracted image extracted as an area corresponding to the certain range from the photographed image, the batch mountain and A background-excluded image generating step for generating a background-excluded image in which the background is excluded from the extracted image in which the background is reflected;
An observation data calculation step of calculating observation data relating to the batch mountain based on the background excluded image. A method for manufacturing a glass article, comprising: