JP3436519B2 - Observation system for plumes - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a volcanic plume observation system for extracting a volcanic plume region and continuously measuring time-series changes in the shape and height of the volcanic plume and the ejection speed.

[0002]

2. Description of the Related Art Fumaroles that are constantly released during non-eruptions in volcanoes include water vapor generated by heating groundwater by heat supplied from magma and volcanic gas originating from magma in the atmosphere. Is caused by being released into. In addition, the fumes from an explosive eruption are a mixture of solid particles originating from magma or solid particles originating from bedrock around the conduit and fumaroles. All of these are emphasized as monitoring items for volcanic activity, as they show the scale of magma underground and its activity in a visible form. In particular, it is qualitatively known that the amount and rate of eruption smoke rapidly increase during an explosive eruption, and it is important to know the change continuously and in real time by transmitting pre-warning information. Helps reduce disasters.

Further, it is considered that the thermal energy, which occupies a major part of the energy emitted from the volcano, and the fumes and fumes have a deep relationship. As a method of estimating the thermal energy supplied by the magma, there are a method of estimating from the total amount of ejected solid matter at the time of eruption and a method of estimating the heat dissipation rate from the highest altitude of smoke. Although a method for easily obtaining the heat dissipation rate from the height of smoke at the time of eruption has been established,
In order to obtain the heat radiation amount, it is necessary to grasp the smoke height in time series. The eruption altitude is often not constant during eruptions, and continuous observation is necessary to understand the eruption altitude. For non-eruption, there is a method to estimate the rising rate of fumarole as an index. In some cases, the rising velocity of fumaroles was obtained by tracking the movement of the feature points of fumaroles, but since the shape of fumaroles is constantly changing, continuous and surface observation is required.

Next, an outline of a conventional observation method for smoke and the like will be described. First, a conventional method for estimating the amount of heat energy released from fumes and fumes will be reviewed. As the contents of the method, there are some research cases on the relationship between the height of smoke and the thermal energy released. For example, Morton et al. (1956) gives the relational expression between the highest smoke height H [m] and the thermal energy release rate Q [watt].

[0005]

[Equation 1] Also, Lucas (1976) gives the following equation in consideration of the influence of wind and turbulence near the ground surface.

[0006]

[Equation 2] These two equations can be applied when the highest altitude can be observed, including solid matter such as plume.

In addition, Briggs (1969) shows that the smoke height h [m] at a point x [m] of the horizontal distance from the heat source.
When there is a wind of μ [m / sec], the following equation is given.

[0008]

[Equation 3] For the relationship between total heat energy and smoke height, see Mor
ton et al. (1956) states that the maximum altitude H [m] and total heat energy Q [J] are expressed by the following equations in the case of standard atmosphere.

[0009]

[Equation 4] These methods use known observations for the highest altitude of smoke. However, the smoke altitude is not observed in time series, and it is not possible to predict when the highest altitude will be observed, so it is not possible to know the heat energy release rate and the total heat energy amount in real time. Moreover, since the time series analysis is not performed, it is not possible to infer the degree of change in magma activity from changes in smoke height and heat dissipation rate. Then, as the existing observation methods, there are the following methods.

(1) Tsuneomi Kagiyama: Thermal energy and H ₂ O emission from fumes from volcanoes (1978, Vol. 2,
(Vol. 23, No. 3) In this method, the height of the fumarine from the reference point is grasped in time series by taking continuous photographs of the fumarine and tracing characteristic points on the screen, and Briggs (1969). ) Is applied to the relational expression between the horizontal distance and height of the fumarolic gas from the heat source at a constant wind speed and the thermal energy release rate to determine the amount of fumarolic heat released during non-eruption.

However, in this method, a still image is manually photographed and the photographing time is controlled by a system which is not linked to the photographing, so that it is necessary to match these. Furthermore, the height of the fumarole is calculated manually from the image. For this reason, observation cannot be performed in real time, and the acquired data cannot be analyzed in time series. In addition, since the feature points of fumarolic gas are extracted with the naked eye from a still image, it is inevitable for the reader to make an error.

(2) Motoharu Jingu and Yukio Ehara: A method for measuring volcanic fumarolic emissions and heat dissipation using the third generation fumarolic diameter (19)
96, Volcano Vol. 41, No. 1) In this method, a model of volcanic fumaroles has a maximum fumarolic center density near the fumarole and a minimum diffusion half-height width, and the central density decreases with time and the half-height width decreases. Assuming that the fumarolic flow rate increases, the maximum diameter and velocity of fumarolic gas are observed by utilizing the fact that the fumarolic flow rate is calculated by the following equation.

[0013]

[Equation 5] In this method, a personal computer was used to analyze the observation results, a video image was read, the flow velocity of the fumarole was measured by the movement of the feature points on the image, and the fumarole diameter was measured from the number of the color of the pixel designated as the fumarole. , The amount of fumarole released is obtained. This analysis system is designed so that an image can be read at an arbitrary sampling time (0.1 to 1 second), and the number of colors of the image can be reduced to 16 for easy analysis.

However, according to this report, it is pointed out that this method is expected to be able to reduce the measurement error by statistical calculation using a large amount of image data of video images. However, such a function is not added to the developed system itself. In addition, this system does not include the image adjustment function required for outdoor observation and the function of performing time-series analysis in real time. Furthermore, the measurement of the fumarolic velocity is based on the extraction of the characteristic points of the image, and the selection of the characteristic points is not automated.

(3) Keiichi Fukui: H ₂ O and thermal energy released from Aso volcano during non-eruption (1995, Volcano No. 4
Volume 0, No. 4) In this method, meteorological conditions (outside air temperature, density, water vapor amount, atmospheric pressure), fumarolic temperature measured by an infrared camera, and a video image obtained by an observation device using a remote sensing method. H ₂ O and thermal energy at the time of non-eruption are calculated from the size and rising speed of the fumarole measured in. In this method, a video image of fumarolic gas is acquired from a remote observation point, the next quantity related to the shape of the plume is obtained, and the ascending velocity is obtained by tracing the clear point of the plume.

However, in this system, the amount relating to the shape of the plume from the image is manually calculated after converting the moving image into a still image. Furthermore, the rising speed is measured manually by selecting the feature points on the image with the naked eye. In addition, this system does not include the image adjustment function required for outdoor observation, the function of performing time-series analysis in real time, and the statistical analysis function.

The current state of the technique for measuring fluids by image analysis is, for example, Toshio Kobayashi, Koji Okamoto et al .: (200
0, Particle Image Velocime
development, visualization information, Vol. 20. No. 7
It is introduced in 7). The quantitative measurement of the flow velocity by the analysis of the visualized two-dimensional image is performed by the particle image velocity measurement method (Particle Image Velocity).
It is called "Etory (PIV)" and is being recognized as a general method due to recent advances in hardware and software. This method takes a means of visualizing a flow and analyzing a digital image of the flow by mixing particles into a flow field, but smoke may be used as particles mixed as a tracer. The algorithm for measuring the flow velocity itself is relatively simple, and some software based on the image correlation method and particle tracking method have been developed.

However, the PIV technique has been tried to be applied to various fields, but at present, it is often used as an industrial measuring technique such as a laboratory experiment for measuring the flow and the shape of the liquid surface of a liquid. For that purpose, high-accuracy image acquisition technology, development of analysis algorithm, development of optimum particle for measurement, development of three-dimensional measurement technology, etc. are being carried out. However, there are very few cases where these technologies are used as a means of observing natural phenomena in the field.

[0019]

Means for Solving the Problems The present invention is to solve the above-mentioned problems and to extract a moving area of outdoor smoke or the like and continuously measure time series changes in height and shape of the smoke or the like. In addition, it is possible to continuously measure the time-series change of the ejection velocity in an arbitrary region such as a plume.

Both the smoke and fumes in the outdoors have a process of rising in the atmosphere as a spirit or a mixture of fine solids and gas. This is plume rise
Called. There are two types of motive force for the plume rise, one is a jet that does not have an initial velocity and rises only by buoyancy due to a density difference, and the other is a jet that does not have a density difference and rises only at an initial velocity. Generally, both fumes and fumes have both of these elements. Since the volcanic plume contains solids and its boundary with the surrounding atmosphere is clear, the moving body can be easily recognized and the maximum altitude can be known. The top of the fumarole is often obscured by diffusion, and the diffused water vapor density distribution cannot be measured directly. However, since it is easy to recognize a water drop in which water vapor is supersaturated as a white cloud, it is possible to approximately understand the region of the spirit.
In the visible light region, the insides of these plumes have a relatively uniform color tone and can be easily compared with the background sky, clouds, and the ground surface. You can use color elements.

The fumes form warm air due to the latent heat accompanying the condensation of water vapor due to the mixing with the inflowing air, in addition to the heat of the ejected gas. The fumes form warm air due to the thermal energy released from the solids that they contain. Therefore, these air masses can be distinguished as a region that is thermally discontinuous with the surrounding atmosphere. From these characteristics, fumes and fumes can be effectively identified using a visualized image such as an optical image in the visible light region or a thermal image in the thermal infrared region.

On the other hand, smoke and fumes generally have a three-dimensional shape and are widely distributed outdoors, and their initial shape is constantly deformed due to the effects of vortices and winds associated with diffusion. The shape is kept for a certain time. Therefore, the feature of the shape can be approximately extracted on a two-dimensional plane having an appropriate cross section. In addition, the shape of details is constantly changing due to convection and diffusion inside the smoke and fumes. Therefore, it is difficult to grasp the whole movement by tracking individual points, but by tracking the movement as a certain area, it is possible to grasp the movement of the whole spirit. Therefore, to observe fumes and fumes,
The method of using a two-dimensional plane image from a remote place to some extent is simple and effective. For these reasons, it is effective to observe smoke and fumes outdoors using continuous images such as video images acquired at remote locations.

To this end, the present invention is an observation system for smoke and the like which extracts a region such as the smoke and continuously measures the time-series changes in the shape and height of the smoke and the ejection speed, and is taken from a fixed point. An image input unit for inputting images continuous in time series, and a pixel-by-pixel luminance difference obtained from a difference image at arbitrary time intervals between the images input by the image input unit are read, and at least time axis integration and scaling are performed. A smoke area extraction unit that fills in holes with a specific brightness portion to extract a smoke area, and a smoke height observation unit that obtains the shape and height of the smoke area extracted by the smoke area extraction unit, and the image. An inter-image correlation analyzing means for detecting an optical flow by an inter-frame correlation calculation at an arbitrary time interval with respect to an arbitrary range of an image input by the input means, and the inter-image correlation solution It is characterized in that a jetting speed observation means for determining the exit velocity from the detected optical flow by means.

The plume shape / height observing means counts the number of detected pixels for each horizontal line of the detected image, determines the plume height from the histogram of the number of detected pixels, and determines the plume height determined as the area of the plume area. It is characterized in that the smoke width is normalized by using it.

The inter-image correlation analysis means normalizes the multiplying coincidence amount of the luminances of the corresponding pixels of the reference image and the unknown image with respect to the average value of the luminances of the entire measurement window region, and calculates the optical flow by the inter-frame correlation calculation. Two-dimensional interpolation is performed by solving simultaneous equations of a quadratic curve that passes through three points, using the adjacent three coefficients that include the maximum coincidence shift position of the image frame correlation coefficient that is detected and calculated in pixel units. , The ejection velocity observation means specifies the reference image in the measurement window and the evaluation function obtains the difference between the shifted position and the most similar position as the velocity or displacement, and measures it. It is characterized in that the ejection vector value is calculated by setting the number of times or measurement time and statistically processing the speed measured a plurality of times.

[0026]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram for explaining an embodiment of an observation system for smoke and the like according to the present invention.
Is an image input unit, 2 is a smoke area extraction unit, 3 is a smoke height observation unit, 4 is an image correlation analysis unit, 5 is a jet velocity observation unit, 6 is an observation data file, and 7 is an observation data output unit.

In FIG. 1, an image input unit 1 is for inputting images which are taken from fixed points and which are continuous in time series. As the images, for example, video images (digital,
Analog), high-sensitivity TV images, infrared thermal images, and still camera images (digital, analog).

The smoke area extraction unit 2 reads the pixel-by-pixel brightness difference obtained from the difference image at arbitrary time intervals, fills in holes by time axis integration and enlargement / reduction, eliminates a specific brightness portion, or the like, that is, the area of the moving body. This is to extract the smoke area. In this way, when discriminating fumes on the image, it is effective to recognize the contour due to the difference in luminance generated in pixel units. In addition, since a moving object such as a volcanic smoke also moves inside, the outline and some moving points are grasped,
By filling the inside of these, the entire moving area can be extracted.

The volcanic plume height observing unit 3 obtains a threshold value that most effectively divides the moving body region obtained by the volcanic plume region extracting unit 2 into a non-moving region by a discriminant analysis method, and determines it as the volcanic plume region. The height of the smoke is determined and the height of the smoke is calculated, and the shape of the smoke is calculated with the smoke width normalized by using the area of the extracted region and the calculated height.

The inter-image correlation analysis unit 4 detects an optical flow by performing an inter-image frame correlation calculation at an arbitrary time interval with respect to an arbitrary range of an image, and the ejection velocity observing unit 5 ejects from the optical flow. The speed is calculated. For volcanic smoke that changes its form continuously, find the most similar form in the image frames at appropriate time intervals and read the change in coordinate values to grasp the direction and speed of movement. You can

The observation data file 6 is a file for storing the data observed by the plume height observation unit 3 and the ejection velocity observation unit 5, and the observation data output unit 7 changes the maximum height of the plume in time series. , The time-series change of the plume shape and the time-series change of the jet velocity of an arbitrary area of the plume are visualized by a list and a real-time graph display, and each data file is output.

In the observation system for smoke and the like according to the present invention, recording of the video image as described above, selection of the difference image,
The reading of the brightness on the image, the determination of the moving area, the calculation of the image correlation, and the display of the analysis result are controlled by the personal computer.

Next, the processing by the observation system for smoke and the like according to the present invention will be described in detail. FIG. 2 is a flow chart of the smoke analysis by the observation system for smoke and the like according to the present invention, FIG. 3 is a diagram showing a setting example of a reference point on the image, a ground target, and an ejection velocity measurement region, and FIG. FIG. 5 is a diagram for explaining recognition, FIG. 5 is a diagram showing an example of a differential image of a grayscale fumigation smoke video image, and FIG. 6 is an example of setting a threshold value by calculation and discrimination analysis of the plume height to prevent misjudgment. FIG. 7 is a diagram for explaining the maximum matching position search and interpolation process, FIG. 8 is a diagram showing a visual display example of the observation result, and FIG. 9 is a view showing the temporal change of the smoke height. It is a figure which shows the converted output example.

In the system for observing volcanic plume according to the present invention,
As shown in FIG. 2, first, set the video image input conditions,
It captures video images of smoke and fumes taken outdoors. When recording, the size of the memory to record,
The number of recording frames, the number of pause frames, the recording time, the number of recording colors, and the recording resolution are set, and after the image is converted into a digital signal, the brightness, contrast, color depth, and hue are appropriately adjusted. In addition, it is necessary to select an appropriate image size in order to measure an object such as fumes and fumes that has a significant change in pattern taken under conditions where observation points cannot be limited outdoors. , The size of the image is selected according to the shooting conditions (step S11).

When measuring the area of smoke and fumes outdoors from a remote location, the observer may not be able to select the most advantageous position for measurement. Therefore, the measurement conditions are set according to the natural conditions of the observation point, taking into consideration the distance to the target object, the surrounding visual field, the noise to be excluded from the measurement, and the like. Then, the area for detecting the area of smoke and fumes is set within the range of the image. Further, since the smoke and fumes rise in the atmosphere, the measurement area is set to a position higher than the arbitrary height within the range of the image (step S1).
2). The position and size of the region a for measuring the ejection velocity are arbitrarily set (step S13). FIG. 3B shows an example of setting the ejection velocity measurement region a on the image.

Further, in order to approximately obtain the size and speed of the detection area in the real coordinates, the conversion method is selected. In the selection of this method, a method of optically obtaining by the lens focal length and a method of geometrically performing the ground orientation from the relationship between the image coordinates and the real coordinates and the coordinates of the image capturing position can be selected. The former has a mechanical error, and the latter has a geometrical error, but the measurement point is not a rigid body and the feature points are not constant.
Considering the size of the area to be obtained, the size and accuracy of the velocity value, these errors can be ignored. Geometrical ground location is performed by setting a two-point ground target (step S14).
FIG. 3A shows an example of setting the reference point and the ground target on the image.

Next, the inter-frame difference processing, the integration in the time axis direction, the filling by enlargement / reduction, the exclusion of the specific luminance portion, and the extraction of the smoke area are performed (steps S15 and S1).
6). A. Extraction of smoke area As shown in FIG. 4, when only the object moves without changing the background, there is a difference image as a method of tracking the object. This method generally assumes that the object is a rigid body whose shape does not change. In the present invention, the fumes and fumes are generally irregular in the outdoors, but they move at a high speed with respect to the background, and it is sufficient to recognize the region as the spirit as a moving body by setting an appropriate time interval. The difference image is used for the extraction of the smoke area by utilizing the possibility.

In the present invention, the brightness information of the image data is used as the characteristic amount for measuring the moving body, and the brightness information is converted from the color RGB image into the grayscale monochrome information.

(1) Difference between frames Extraction of a smoke area basically uses difference information of images of two frames. At time t, the brightness at the position (x, y) is I (t,
x, y), two times t = tl and t = t2 (t2>
Difference information D (x, y) is created using the image of tl).

[0040]

[Equation 6] A fixed threshold th is set, and when the value of D (x, y) is larger than th, it is set as a moving area. If there is a clear pattern at all positions in the smoke area, this process can extract the smoke area, but in the case of a video image, the brightness range that can be captured is limited, so the smoke area is There are areas where information is missing. Therefore, only a part of the outline of the smoke area is often extracted only by performing the difference processing. In addition, it is unavoidable that noise on the image is recognized as a brightness difference and erroneous detection occurs. FIG. 5A shows an example of a grayscale fumes smoke video image, and FIG. 5B shows an example of a difference image thereof.

(2) Integration in the time axis direction In order to extract the entire contour of the smoke area, the detection area is integrated in the time axis direction. In this time axis integration,
When a is a constant of 1 or less, and the difference information of the brightness difference one frame before, which is multiplied by a, is accumulated, the past difference information is gradually attenuated and eventually has no effect. It is outside the detection range after image sampling.

(3) Filling by enlargement / reduction After the integration in the time axis direction is performed using the difference between the frames, it is necessary to fill the inside of the contour in order to extract the entire smoke area. The scaling method is a general image measurement method that unifies the separated shapes into one category and fills in a region that is partially lost by binarization. In this process, the detection points located at positions distant from each other in the original image are integrated, and the periphery of the combined result is scraped.

First, temporary storage areas T1 (x, y) and T2 (x, y) having the same size as the image size are secured. Tl
All are initialized to "non-detection" and all T2 are initialized to "detection". The resulting image is then examined pixel by pixel and separated into "detected" and "non-detected" pixels. All the areas having the radius rl designated with the "detection" pixel as the center are registered as the detection areas in the temporary storage area. Furthermore, the temporary storage area Tl is investigated. Regarding the position of the "non-detection" pixel, all the areas having the radius r2 in the temporary storage area T2 are set as non-detection areas.
The temporary storage area T2 obtained as a result is set as the final result.
In the original image, distant detection points within 2 × rl can be combined.

(4) Exclusion of Specific Luminance Part When separating the smoke area by brightness information, the smoke often contains a large amount of white water vapor outdoors, so that it can be clearly separated from a low brightness area such as the ground. To extract the ground brightness, find the average and variance of the ground brightness,
It is considered that there is an eruption smoke in a region with an appropriate luminance value or more. Also,
The brightness of the sky and the brightness of the smoke area are also different. Since the brightness of the smoke area may be higher or lower than the brightness of the sky, not only the size comparison as when the ground area is excluded, but an appropriate range is specified and excluded.

(5) Detection of smoke height The height of smoke should be determined after eliminating false detection points or false detection areas isolated from the smoke area as much as possible. In the smoke area, assuming that the width of the smoke is relatively constant, count the number of detected pixels for each horizontal line of the detected image,
A method is conceivable in which the number of pixels that most effectively divides the histogram of the number of detected pixels into an “emission” region and a “non-detection” region is determined and the smoke height is determined from the detected image based on the number of pixels.

Considering a certain horizontal line y, when the horizontal line is included in the smoke area, it is expected that the total sum of the number of detected pixels will be close to a certain value depending on the width of the smoke. Further, when it is not included in the smoke area, it is expected to be close to a certain value (0) outside the smoke area. Therefore, if the threshold for dividing the width of the smoke area and 0 into the two can be obtained most effectively, the smoke height can be obtained by sequentially searching from the upper part of the detected image and finding the horizontal line that exceeds the threshold for the first time. it can.

To obtain the threshold value, the method of Otsu is used. This is a method of obtaining the most effective threshold value when binarizing a grayscale image. A discriminant analysis method is used to search for a threshold value that maximizes the inter-class distance between two groups when the two groups are separated by the brightness of the image.

(6) Calculation of normalized plume width It is difficult to describe the shape of plume accurately. In existing research reports, there is an example in which the three-dimensional shape of plume is measured as an ellipsoid or an inverted cone. In the case of observing fumes and fumaroles using the present invention, the direction of the fumes varies depending on the ejection conditions and atmospheric weather conditions, and therefore, it is not always possible to observe a cross-sectional shape whose volume can be calculated. Therefore, in the present invention, the shape of the plume is expressed by a two-dimensional cross section, and the value obtained by dividing the detected plume area by the plume height is displayed as the normalized plume width. This value is used to simply discriminate between broad and elongated plumes.

The ejection speed of fumes and fumes is measured (step S17). B. Measurement of ejection velocity of fumes and fumes To obtain ejection velocity of fumes and fumes, velocity vectors are calculated using adjacent frames of the image (step S1).
8). There are two methods of obtaining the movement vector (optical flow) of an image: a method of tracking a feature point of a moving body and a method of shifting two images to obtain a planar image frame correlation. The shape of smoke and fumes changes continuously and smoothly as a moving body. However, the shape is constantly changing, and is not changing uniformly throughout. Therefore, tracking a certain feature point does not always represent the movement amount of the area. In the method of obtaining the correlation between image frames, the amount of displacement of the object position can be obtained by calculating the cross-correlation function for various displacements and searching for the displacement that gives the maximum value of the results. Therefore, in the present invention, a method of obtaining the correlation between image frames is used. Further, the image correlation function is calculated by using the method of obtaining the level matching area after binarizing the input image.

(1) Correlation calculation between image frames In order to calculate the correlation between image frames, a function for evaluating how similar the reference image (template) and the unknown image are to each other is required. The evaluation function used at this time generally includes “sum of brightness differences” and “normalized correlation”. These calculation amounts are significantly different, and the “sum of brightness differences” has a shorter calculation time. However, when a displacement vector such as a volcanic smoke is obtained outdoors, if the brightness of the image differs with the frame, the vector may not be accurately detected by the "sum of brightness differences". Therefore, in the present invention, the normalized mutual operation is adopted.

The "normalized correlation" is a value obtained by normalizing the multiplying coincidence amount of the luminances of the corresponding pixels of the reference image and the unknown image with respect to the average luminance value of the entire measurement window area.
It is "1" when they completely match, and "0" when they are uncorrelated (such as two images having no relation).

In measuring the velocity / displacement vector, the reference image is designated in the set measurement window. In order to detect in which direction and how much this area has moved in the next frame, the position of the measurement window is gradually shifted around the reference image, and the evaluation function is used to calculate the “similarity” to each displacement position. "The degree that it has". The difference from the most similar position is defined as the speed or displacement to be obtained.

(2) Sub-pixel positing When remotely observing smoke, fumes, etc. outdoors, the pixel resolution of the digitized image is limited, so obtaining the optical flow on a pixel-by-pixel basis will provide sufficient correlation. I can't. In the present invention, the adjacent 3 coefficients including the maximum value (maximum coincidence shift position) of the inter-image frame correlation coefficient obtained in pixel units as shown in FIG. 7A are used, and FIG. The two-dimensional interpolation is performed by solving a simultaneous equation of a quadratic curve that passes through three points as shown in (1) to obtain a correlation with an accuracy of a pixel unit or less.

(3) Examination of reliability In this system, the reliability of the measurement data and the suitability of the used image are confirmed with respect to the measurement result. Regarding the similarity between the frames used for the measurement, the Pk value (normalized peak height) is used in which the correlation function value at the subpixel maximum matching position is normalized by the measurement window area and corrected to 100% at the time of perfect matching. . For the clarity of the maximum matching position, use the values around the maximum matching position.
The determination is made using a value that represents the sharpness of the origin peak of the correlation function when the next interpolation is performed. These values are output as a data list together with the velocity vector measurement values.

(4) Statistical processing of data, calculation of jet velocity vector Since jets of smoke and fumes are not always steady outdoors, the jet velocity vector is measured a plurality of times by setting the number of times of measurement and the measuring time, Statistical processing for obtaining the average value and variance is performed, and the jet vector value subjected to the statistical processing is calculated. C. Display of results, output of data The fumes, the height of the fumes, the normalized fumes width, and the ejection speed measured by performing the above-described processing are converted into an actual distance and output (steps S19 and S20). In the present invention, the observation results are visualized by displaying these results in a list and a simple graph. In addition, the measurement results described in time series are displayed in real time and are output as a data file, and the ejection velocity vector is output as a data file for each measurement before statistical processing. These data files will be used for more sophisticated analysis of the observations. FIG. 8A shows a display example of the velocity vector in the set measurement region, and FIG. 8B shows a display example in which the ejection velocity vector is relatively small. Further, FIG. 9 shows an output example in which the temporal change in the height of the smoke is visualized.

[0056]

As is apparent from the above description, according to the present invention, the following effects can be obtained.

It is possible to measure smoke or fumes as a natural phenomenon outdoors in real time. That is, the tracking of a moving body and the analysis of the motion form of a fluid using a video image have been widely used in recent years in a field such as a laboratory experiment. However, there are few systems for directly observing smoke and fumes as natural phenomena outdoors using moving images, and conventional outdoor observation equipment using moving images also
Only the image recording device was installed for the purpose of monitoring. According to the present invention, it is possible to consistently record a moving image, process an image, and output an analysis result under the control of a personal computer, and acquire an image under an outdoor condition to perform an analysis system. In addition to having the function of setting the position information of the observation point and adjusting the measurement range in order to perform it in real time, it is possible to easily set the measurement conditions and analysis conditions necessary for performing the analysis.

By image processing, the form of fumes and fumes can be measured continuously, automatically and quantitatively. In other words, the conventional systems do not use image processing algorithms for measuring the moving images of fumes and fumes acquired by a video camera or the like on those images. According to the present invention, a moving image is read from a difference image at arbitrary time intervals by using a moving image, and the moving object is read by an image processing algorithm such as time-axis integration or hole filling by enlargement / reduction or exclusion of a specific luminance portion. Area can be extracted. Furthermore, the ejection velocity can be calculated by using an image processing algorithm that detects an optical flow in an arbitrary range of an image by an inter-image correlation calculation at an arbitrary time interval. As a result, the characteristic of the present invention is that the measurement can be automated and the continuous measurement in time series over a long period of time can be quantitatively and easily performed.

It is possible to output the observation result according to the analysis means. That is, in the conventional smoke and fumarole observation systems, it has been common to take the measurement results out of the system as raw data for analysis. According to the present invention, in accordance with the purpose of analysis, continuous data is acquired in a time series at arbitrary time intervals, and the results of measurement by image processing are visualized in the form of graphs, lists, data files, etc. in real time. Can be output. Furthermore, there is a great feature in that these data can be analyzed, statistically processed and output.

As described above, according to the present invention, the volcanic smoke of a volcano or the like is continuously monitored, and the amount of volcanic smoke, the reaching altitude, and the jet speed near the jet outlet are quantitatively and continuously automatically measured. be able to. It is considered that the amount and rate of eruption smoke are closely related to the activity of underground magma, and it has already been performed to estimate the amount of heat radiation from the rate of eruption and the diameter of smoke plume. The present invention continuously changes the velocity of smoke from a remote point,
By automatically observing, statistical analysis of measured values can be easily performed. Furthermore, since it is possible to make a quantitative and real-time comparison of the eruption velocity with the amount of smoke and the height of smoke, it is possible to quantitatively and efficiently monitor volcanic activity using images that are qualitatively visually observed by humans. It can be carried out. The observer can also read a large change from the measurement values continuously obtained by this system and issue an alarm. In addition, this system is useful as an observation instrument for basic research because it can deposit the internal structure and development process of the plume by measuring the velocity distribution of each part of the plume.

The present invention is not limited to the above embodiment, and various modifications can be made. For example, in the above-mentioned embodiment, the explanation was made assuming that the observation of fumes and fumaroles was performed, but as fumes, etc., not only fumes, fumes but also clouds and fog, factory smoke, avalanche behavior analysis, fire monitoring, etc. It can also be used to observe various phenomena. Therefore, it can be applied to, for example, environmental fields such as local cloud / fog generation and factory smoke emission monitoring, snow and ice fields such as avalanche behavior analysis, and fire fighting fields such as fire monitoring. Furthermore, it tracks moving objects that have irregular contours such as slope collapse and landslides and that move in a certain direction.
It can also be applied to extract a region and measure a moving speed.

[Brief description of drawings]

FIG. 1 is a diagram for explaining an embodiment of an observation system for smoke and the like according to the present invention.

FIG. 2 is a flow chart of a volcanic plume analysis by a volcanic plume observation system according to the present invention.

FIG. 3 is a diagram showing a setting example of a reference point, a ground target, and an ejection velocity measurement region on an image.

FIG. 4 is a diagram for explaining recognition of a moving area based on a difference image.

FIG. 5 is a diagram illustrating an example of a difference image of a grayscale fumigation smoke video image.

FIG. 6 is a diagram for explaining an example of setting a threshold value by calculation of smoke height and discriminant analysis to prevent erroneous discrimination.

FIG. 7 is a diagram for explaining a search for a maximum matching position and an interpolation process.

FIG. 8 is a diagram showing an example of a visualized display of observation results.

FIG. 9 is a diagram showing an output example in which a temporal change in the smoke height is visualized.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Image input part, 2 ... Smoke area extraction part, 3 ... Smoke height observation part, 4 ... Image correlation analysis part, 5 ... Ejection velocity observation part, 6
… Observation data file, 7… Observation data output section

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ Identification code FI G06T 7/20 G01C 15/00 104Z // G01C 15/00 104 G01B 11/24 K (56) Reference JP-A-6-20052 (JP, A) JP-A-6-308256 (JP, A) Masaki Ikegami, Kunio Onda, Yoshinao Aoki, “Detection Method of Flowing Powder (Snow) by Image”, IEICE Technical Report, Japan, Institute of Electronics, Information and Communication Engineers, February 23, 1990, Vol. 89, No. 436, p. 23-28 Keiichi Fukui, "H2O and Thermal Energy Emitted from Non-Eruption from Aso Volcano", Hiyama , Japan, Volcano Society of Japan, September 20, 1995, Volume 40, No. 4, p. 233-248 Tsukasa Jingu, Yukio Ehara, "Method for measuring volcanic fumarolic emissions and heat radiation using maximum fumarolic diameter," Volcano, Japan, Volcano Society of Japan, March 5, 1996, Vol. 41, No. 1. , p. 23-29 Masahiro Muikaichi, Nozomu Hamada, “Optical Flow Estimation Considering Velocity Constrained Equation and Edge Connection at Edge”, IEICE Transactions, Japan, The Institute of Electronics, Information and Communication Engineers, 1999 3 25th, Vol. J82-D-2, No. 3, p. 407-414 Kazunobu Muraoka, Norishige Chiba, Hiromichi Takahashi, Mamoru Miura, “Two-dimensional CG simulation of smoke and flame by eddy field and particles”, IEICE Transactions, Japan, The Institute of Electronics, Information and Communication Engineers, 1993 August 25, J76-D-2, No. 8, p. 1735-1745 Toshio Kobayashi, Koji Okamoto, Tetsuo Saga, "Development of the Particle Image Velocity", Visualization Information, Japan, Visual Information Society of Japan, April 1, 2000, Volume 20, No. 77, p. 107-112 Kaneyama, Tsuneomi, "Heat energy and H2O emission from volcanic fumes-Estimated from plume rise", Volcano, Japan, Volcano Society of Japan, October 1, 1978, Volume 23, Vol. No. 3, p. 183-197 (58) Fields surveyed (Int.Cl. ⁷ , DB name) G01W 1/00-1/18 G01B 11/00-11/30 G01P 3/00-3/80 G06T 1/00-1 / 60 G06T 7/00-7/60 G01C 15/00-15/14 JISST file (JOIS)

Claims (7)

(57) [Claims] 1. A region of eruption smoke or the like is extracted to determine the shape of eruption smoke or the like.
An observation system for volcanic plume, etc. for continuously measuring time-series changes in height and ejection speed, comprising image input means for inputting time-sequentially continuous images taken from a fixed point, and input from the image input means. Read the brightness difference in pixel units obtained from the difference image at any time interval of the image,
Fume area extraction means for extracting at least a time axis integration, hole filling by expansion / reduction, and removal of a specific brightness part to extract a smoke area, and a smoke height / shape for obtaining the smoke shape / height of the smoke area extracted by the smoke area extraction means. Observing means, an inter-image correlation analyzing means for detecting an optical flow by an inter-frame correlation calculation at an arbitrary time interval for an arbitrary range of an image inputted by the image input means, and the inter-image correlation analyzing means A system for observing smoke and the like, comprising: a jet velocity observing means for obtaining a jet velocity from the detected optical flow. 2. The plume shape / height observing means counts the number of detected pixels for each horizontal line of the detected image and determines the plume height from the histogram of the number of detected pixels. Etc. observation system. 3. The plume shape / height observing means calculates a plume width normalized by using the plume height determined to be the area of the plume region, to observe the plume smoke and the like. system. 4. The inter-image correlation analysis means normalizes the multiplication and coincidence amount of the luminances of the corresponding pixels of the reference image and the unknown image with respect to the average value of the luminances of the entire measurement window region, and performs the optical correlation by the image frame correlation calculation. The system for observing plume smoke according to claim 1, wherein a flow is detected. 5. The inter-image correlation analysis means uses three adjacent coefficients including the maximum coincidence shift position of the inter-image frame correlation coefficients obtained in pixel units in the center, and simultaneous equations of quadratic curves passing through three points. 5. The system for observing volcanic smoke and the like according to claim 4, wherein a two-dimensional interpolation is performed by solving the above to obtain a correlation with an accuracy of pixel unit or less. 6. The ejection velocity observing means specifies a reference image in a measurement window and obtains a difference between each shifted position and the most similar position as a velocity or displacement by an evaluation function. Observation system for smoke from Japan. 7. The eruption smoke value according to claim 6, wherein the ejection velocity observing means calculates the ejection vector value by statistically processing the velocity measured a plurality of times by setting the number of times of measurement or measurement time. Observation system.