WO2021235155A1

WO2021235155A1 - Cloud location calculation system, method, and program

Info

Publication number: WO2021235155A1
Application number: PCT/JP2021/016017
Authority: WO
Inventors: 祐弥 ▲高▼島
Original assignee: 古野電気株式会社
Priority date: 2020-05-20
Filing date: 2021-04-20
Publication date: 2021-11-25
Also published as: JPWO2021235155A1

Abstract

In order to provide a cloud location calculation system, method, and program which can calculate location information about clouds for each mass of clouds, the system (1) comprises: an image acquisition unit (11) which acquires sky images (G1–G3) that include at least the sky and are captured by a plurality of cameras (10) disposed at different locations; a first cloud location calculation unit (12) which calculates location information about clouds in a first area (Ar1) on the basis of a comparison between sky images that are among the plurality of sky images (G1-G3) and are acquired by at least two cameras (10); an extraction unit (13) which extracts masses of clouds (m1-m5) from the location information about the clouds in the first area (Ar1); a second area setting unit (14) which sets a second area (Ar2) that includes the extracted masses of clouds (m1-m5) and is narrower than the first area (Ar1); and a second cloud location calculation unit (15) which calculates location information about a mass of clouds in the second area (Ar2) on the basis of a comparison between images corresponding to the second area (Ar2) in the sky images that are among the plurality of sky images (G1-G3) and are acquired from at least two cameras (10).

Description

Cloud position calculation system, method, and program

This disclosure relates to cloud position calculation systems, methods, and programs.

It is known to use an omnidirectional camera installed on the ground as a cloud observation method. For example, in Patent Document 1, the sky is imaged with one spherical camera, all-sky images at different times are acquired, and the movement of the same cloud reflected in the all-sky image is tracked, so that the speed of the cloud and the cloud flow. There is a statement that the direction will be decided.

Patent Document 2 describes that two spherical cameras are used to calculate the altitude of clouds vertically above equipment whose distance from the two spherical cameras is known.

Jitsukaisho 57-160681 Gazette Japanese Unexamined Patent Publication No. 2012-242322

However, in the method of Patent Document 1, the altitude of the cloud is unknown, so the altitude is tentatively determined. The method of Patent Document 2 seems to be able to calculate only the altitude of the cloud directly above the place where the distance of the camera is known. Therefore, at present, no method has been proposed for calculating the position information (altitude, latitude / longitude) for each cloud for a plurality of clouds shown in the image.

The present disclosure provides a cloud position calculation system, a method, and a program capable of calculating cloud position information for each cloud mass.

The cloud position calculation system of the present disclosure is from an image acquisition unit that acquires at least an aerial image including the sky taken by a plurality of cameras arranged at different positions from each other, and at least two of the plurality of aerial images. A first cloud position calculation unit that calculates cloud position information in the first region based on the comparison of the acquired sky images, and an extraction unit that extracts cloud masses from the cloud position information in the first region. The second area setting unit that includes the extracted cloud mass and sets a second area narrower than the first area, and the first in the sky image acquired from at least two cameras among the plurality of sky images. It is provided with a second cloud position calculation unit that calculates the position information of the cloud mass in the second area based on the comparison of the images corresponding to the two areas.

In this way, the position information of the clouds in the relatively wide first area is calculated based on the comparison of the aerial images, and the second area narrower than the first area including the cloud mass extracted from the position information of the clouds is targeted. Since the position information of the cloud mass is calculated by comparing the sky image, the position information can be calculated accurately for each cloud mass compared to the case where the comparison is performed by focusing on a part of the cloud mass of the sky image from the beginning. Become. Nevertheless, first, since the position information of the clouds in the relatively wide first region is calculated, the consistency as a whole can be ensured, and then the position information of the local cloud mass in the second region can be reduced in estimation error. It becomes possible.

A block diagram showing the configuration of a cloud position calculation system. A flowchart showing the processing executed by the cloud position calculation system. A flowchart showing the processing executed by the cloud position calculation system. Explanatory drawing about setting of first area. The figure which shows the relationship between the 2nd region and the 3rd region included in the 1st region. The figure which shows the relationship between the aerial image and the third area. Explanatory diagram about conversion of clouds to cloud distribution matrix in the third region. The figure which shows the process of calculating the position information of the cloud in the 3rd region based on the contrast of the aerial image at the altitude h = 1325m. The figure which shows the process of calculating the position information of the cloud in the 3rd region based on the contrast of the aerial image at the altitude h = 2575m. The figure which shows the process of calculating the position information of the cloud in the 3rd region based on the contrast of the aerial image at the altitude h = 3575m. The figure which shows the process of calculating the position information of the cloud in the 3rd region based on the contrast of the aerial image at the altitude h = 1800 m. The figure which shows the process of calculating the cloud distribution matrix in the 3rd region and the 4th region from the sky image obtained from the camera installation point, and obtaining the position information of the cloud in the 1st region. The figure which shows the process of extracting a cloud mass from the position information of a cloud in the 1st area. The figure which shows the cloud mass in the 1st region and the 2nd region. The figure which shows the calculation result of the altitude of each cloud mass included in the 1st area.

Hereinafter, the cloud position calculation system of the present disclosure will be described with reference to the drawings.

The cloud position calculation system 1 is used for the observation system. The system 1 includes two or more cameras 10 that capture the sky and a computer that processes the sky images taken by the cameras 10. The system 1 acquires sky images acquired from at least two cameras 10 arranged at different positions from each other, and calculates unknown cloud position information based on the images. In the present specification, the cloud position information includes at least one of the latitude / longitude information and the altitude information of the cloud. The camera 10 may be any camera as long as it can photograph the sky. In this embodiment, an omnidirectional camera using a fisheye lens is installed facing upward in order to capture a wide area of the sky with one camera. If the camera 10 is installed so as to be vertically upward and horizontal, the center of the aerial image obtained from the camera 10 is directly above (elevation angle 90 °), and the elevation angle becomes smaller toward the edge of the image from the center. Further, if the orientation of the camera 10 is known, the orientation in the image is also known. That is, each pixel at an arbitrary position in the aerial image has a known elevation angle and a known orientation.

In this embodiment, as shown in FIG. 4, an example in which a plurality of cameras 10 are installed at different positions in Hachijojima, Tokyo, Japan, clouds are observed, and cloud position information is calculated will be described. Each of the plurality of cameras 10 is installed at the camera installation points C1, C2, C3, and C4 shown in FIG.

As shown in FIG. 1, the system 1 includes an image acquisition unit 11, a first cloud position calculation unit 12, an extraction unit 13, a second area setting unit 14, and a second cloud position calculation unit 15. .. In each of these parts 11 to 15, software and hardware cooperate by executing a program in which the processor 1b is stored in the memory in advance in a computer equipped with a processor 1b such as a CPU, a storage 1a such as a memory, and various interfaces. Will be realized.

<Image acquisition unit 11>
As shown in FIGS. 4 and 12, the image acquisition unit 11 acquires aerial images G1, G2, and G3 including at least the sky taken by a plurality of cameras 10 installed at different positions. In this embodiment, as shown in FIG. 4, four aerial images are acquired from each camera installation point C1, C2, C3, C4. These four aerial images are taken at the same time by synchronizing the time of the camera 10 in order to contrast with each other. The time error is within 1 second. It is preferable that the time is the same as described above, but it is not necessary that the time is exactly the same. A plurality of aerial images may be taken at times corresponding to each other. For example, it may be in the same time zone. The same time zone refers to, for example, the same time in a state where the times of a plurality of cameras are synchronized with each other. However, in the field in which the present invention is carried out, in order to compare the results of photographing a very wide range, it is preferable that the time difference between the cameras pointed to by the same time zone is zero or small, but the time difference is, for example, about 10 minutes. It is possible to exert the effect even if there is. It is more preferably within 2 minutes, preferably within 1 minute, and more preferably within 10 seconds.

<First cloud position calculation unit 12>
The first cloud position calculation unit 12 calculates the cloud position information in the first region Ar1 based on the comparison of the sky images acquired from at least two cameras among the plurality of sky images. The first region Ar1 is a geographical region based on latitude and longitude, as shown in FIGS. 4 and 14, and corresponds to, for example, a certain region on a map. In the examples of FIGS. 4 and 14, the first area Ar1 is set to include Hachijojima. The first area Ar1 can be changed as appropriate. 13 and 14 show cloud position information (latitude / longitude and its distribution) in the first area Ar1 calculated by the first cloud position calculation unit 12. The cloud position information in the first region Ar1 indicates the latitude / longitude, its distribution, and the altitude of the cloud. The altitudes of the clouds are all the same. In the examples shown in FIGS. 4 to 11, three aerial images acquired from the three cameras 10 are used, but the present invention is not limited thereto. The number of cameras may be two or more. When there are two cameras, the cloud position information is calculated by comparing the two sky images acquired from the two cameras. That is, if the number of cameras is two or more, the number can be changed to any number, and the combination of two or more aerial images can be changed arbitrarily.

A specific description of the configuration and operation of the first cloud position calculation unit 12 will be described later. As long as the cloud position information in the first region Ar1 can be calculated, the present invention is not limited to this embodiment. The first cloud position calculation unit 12 shown in FIG. 1 includes a common area setting unit 12a, a subregion setting unit 12b, a subregion cloud position calculation unit 12c, and a synthesis unit 12d.

The common area setting unit 12a shown in FIG. 1 is a common area Ar1 used to combine a plurality of areas specified by a set of aerial images. In the present embodiment, the common area is used as the first area Ar1, but the common area and the first area Ar1 do not have to be the same. As shown in FIG. 4, the common area setting unit 12a sets the geographical area as the common area Ar1 in order to generate the position information of the clouds. In the example of FIG. 4, the common area Ar1 is set based on the maximum value and the minimum value of the latitude and longitude of each camera installation point C1 to C4. Specifically, the predetermined distance D1 is wider than the maximum and minimum values of latitude and longitude, which is an example. In the example of FIG. 4, D1 = 5000 m, and the common area Ar1 is separated by a mesh of 50 m. Two matrices corresponding to each cell of the mesh of the common area Ar1 were generated, and a cloud distribution matrix showing the distribution of clouds and a cloud altitude matrix showing the altitude of the bottom of the clouds were generated.

The small area setting unit 12b shown in FIG. 1 generates all sets of two cameras 10 from all the cameras 10 (C1 to C4), and sets the small area Ar10 for each set. The subregion Ar10 is used to represent the position information of the clouds obtained by contrasting the aerial images obtained from the two corresponding cameras 10. In the present embodiment, as shown in FIG. 5, as the small area Ar10 set by the small area setting unit 12b, the third area Ar13 and the fourth area Ar14 will be described as an example. The third region Ar13 and the fourth region Ar14 are each included in the first region Ar1. As shown in FIG. 12, the third region Ar13 shown in FIG. 5 is the cloud in the third region Ar13 based on the comparison of the aerial images G1 and G2 acquired from the cameras 10 installed at the camera installation points C1 and C2. Set to calculate location information. As shown in FIG. 12, the fourth region Ar14 shown in FIG. 5 is the cloud in the fourth region Ar14 based on the comparison of the aerial images G2 and G3 acquired from the cameras 10 arranged at the camera installation points C2 and C3. Set to calculate location information. In the present embodiment, each subregion Ar10 is set based on the center of the two camera installation points, but the setting method can be changed as appropriate. In order to facilitate the synthesis process described later, each subregion Ar10 is generated by cutting out from the common region Ar1.

The subregion cloud position calculation unit 12c shown in FIG. 1 captures cloud position information in the set subregion Ar10 (third region Ar13, fourth region Ar14) from aerial images acquired from the corresponding two cameras 10. Calculated based on the comparison of. Hereinafter, a method of calculating the position information of clouds in the third region Ar13 based on the aerial images G1 and G2 will be described.

FIG. 6 is a diagram showing the relationship between the aerial images G1 and G2 and the third region Ar13. In order to clearly show the positional relationship between the coordinates in the third region Ar13 and the respective sky images G1 and G2, the points are plotted so as to form a cross. Each camera 10 has a known latitude and longitude. Further, all the pixels in the aerial images G1 and G2 have known elevation angles and azimuth angles, respectively. Therefore, any pixel in the aerial image can be associated with the latitude and longitude in the third region Ar13. Specific examples are shown as the following two (1) and (2).

(1) The elevation angle α1 and the azimuth angle (for example, the angle indicating the direction with the north as 0 degree) are known for the pixel P1'(the cloud is reflected) in the sky image G1 shown in FIG. If the altitude h is tentatively determined, the distance L1 from the point P1 on the ground surface where P1'is projected directly below to the camera installation point C1 can be calculated. The latitude / longitude of the point P1 can be calculated based on the calculated distance L1, the known azimuth, and the latitude / longitude of the camera installation point C1, and this is the latitude / longitude of the cloud.

(2) In order to know the pixel P2'in the aerial image G2 corresponding to the point P2 in the third region Ar13 shown in FIG. 6, the distance L2 and the azimuth of the point P2 with respect to the camera installation point C2 are known. Therefore, if the altitude h is tentatively determined, the elevation angle α2 with respect to the point P2'at the altitude h directly above the point P2 on the ground surface can be calculated. Based on the calculated elevation angle α2 and azimuth angle, the pixel P2'in the aerial image G2 can be specified.

The small area cloud position calculation unit 12c uses the correspondence between the sky image and the third area Ar13 to obtain the cloud distribution on the third area Ar13 based on the latitude and longitude, as shown in FIG. into a cloud distribution matrix a ₃ and cloud height matrix (not shown) indicate. Cloud distribution matrix A ₃ inputs 1 to the value of the element corresponding to the cells of the mesh cloud exists clearly, 0.5 is input to the value of the element corresponding to the cell of the mesh thin clouds are present There is. Of cloud altitude matrix, the elements corresponding to the elements that there are clouds in cloud distribution matrix A _3, the value of the altitude h of the temporarily determined is input. Of cloud distribution matrix A ₃ and cloud altitude matrix, the values of elements that do not exist cloud becomes zero. The value of the elements of the cloud distribution matrix _{A 3} is 0.0 or more and take 1.0. The value of the element is also called cloud cover.

The small area cloud position calculation unit 12c tentatively determines the altitude h, generates the cloud distribution matrix _{A3_G1} and the cloud altitude matrix of the third area Ar13 based on the sky image G1 as shown in FIG. 12, and generates the sky image G2. The cloud distribution matrix _{A3_G2} and the cloud altitude matrix of the third region Ar13 are generated based on the above. Next, the subregion cloud position calculation unit 12c calculates the degree of coincidence (reliability) between the _{cloud distribution matrices A 3_G1} and A _{3_G2.} For the reliability, an evaluation function ZNCC (Zero-mean Normalized Cross-Correlation) representing the similarity between two matrices having a value of 0.0 or more and 1.0 or less was used, but the evaluation function is not limited to this. .. Of course, pattern matching of another method may be used.

The subregion cloud position calculation unit 12c calculates the cloud distribution matrices A _{3_G1} , A _{3_G2} , the cloud altitude matrix, and the reliability ZNCC, respectively, by making the altitude h different by a predetermined altitude. When the search for the altitude in the predetermined range is completed, the cloud distribution matrix and the cloud altitude matrix corresponding to the highest reliability ZNCC are adopted, and as shown in FIG. 12, the reliability ZNCC _{3_FIX} , the cloud distribution matrix A _{3_FIX,} and the cloud altitude matrix are adopted. Is stored in the storage unit. A similar process is performed for all subregions Ar10. _{In FIG. 12, a plurality of sets of cloud distribution matrices A 4_G2} , _{A4_G3} , cloud altitude matrix and reliability ZNCC are calculated based on the comparison of aerial images G2 and G3 for the fourth region Ar14, _{and the maximum reliability ZNCC 4_FIX} and clouds are calculated. The distribution matrix A _{4_FIX} and the cloud altitude matrix are stored in the storage unit.

As shown in FIG. 12, the synthesis unit 12d shown in FIG. 1 obtains cloud position information of each subregion Ar10 (including the third region Ar13 and the fourth region Ar14) calculated by the subregion cloud position calculation unit 12c. Synthesize to the common area Ar1. In the example shown in FIG. 12, the cloud position information in the third region Ar13 and the cloud position information in the fourth region Ar14 are combined to calculate the cloud position information in the first region Ar1. Here, the region where the third region Ar13 and fourth regions Ar14 do not overlap, may be adopted the value of the corresponding elements of each cloud distribution matrix A _3, A ₄ as it is. For the area where the third area Ar13 and the fourth area Ar14 overlap, the cloud position information at the overlapping point, the cloud position information of the corresponding third area Ar13 and the fourth area Ar14, and the reliability ZNCC of each. It is preferable to calculate based on the weights of _{3_FIX} and ZNCC _{4_FIX.} For example, the value A (i, j) of the element of the cloud distribution matrix A at the overlapping portion can be calculated by the following equation.
A (i, j) = A _{3_FIX} (i, j) x ZNCC _{3_FIX} + A _{4_FIX} (i, j) x ZNCC _{4_FIX}
i and j are row numbers and column numbers of a matrix indicating overlapping points.

8 to 11 are diagrams showing a process of calculating the position information of clouds in the third region Ar13 based on the comparison of the aerial images G1 and G2. FIG. 8 shows aerial images G1 and G2 at an altitude of h = 1325 m, cloud distribution matrices _{A3_G1} and _{A3_G2,} and a plot of reliability ZNCC. FIG. 9 is a diagram corresponding to FIG. 8 when the altitude h = 2575 m. FIG. 10 is a diagram corresponding to FIG. 8 when the altitude h = 3575 m. FIG. 11 is a diagram corresponding to FIG. 8 when the altitude h = 1800 m. The shooting point A in the figure is the camera installation point C2, and the shooting point B is the camera installation point C1. It can be understood that the altitude h = 1800 m shown in FIG. 11 has the highest reliability ZNCC, and the cloud altitude h = 1800 m.

<Extraction unit 13>
As shown in FIG. 13, the extraction unit 13 shown in FIG. 1 extracts cloud masses (m1 to m5) from the position information of clouds in the first region Ar1. In this embodiment, as shown in FIG. 13, each element of the cloud distribution matrix in the first region Ar1 has a numerical value indicating the existence of a cloud. In FIG. 13, the elements having a high existence probability are shown in white, and the light elements are shown in gray. Although the altitude of each cloud in the first region Ar1 is the same, the altitude is actually different for each cloud mass, so the extraction unit 13 extracts the cloud mass. As an example, the extraction unit 13 extracted the elements whose cloud distribution matrix elements have a predetermined value or more, and grouped them by a labeling algorithm used for general image processing. In FIG. 13, if even one of them is adjacent to each other in the 8 neighborhood search, it is determined that they are the same cloud mass. The search is not limited to 8 neighborhoods, and 4 neighborhoods may be used. In the example of FIG. 13, five cloud masses m1 to m5 are shown as an example.

<Second area setting unit 14>
The second area setting unit 14 shown in FIG. 1 sets the second areas Ar21 to Ar25 including the cloud mass extracted by the extraction unit 13. The second region Ar2 may be narrower than the first region Ar1 and may include a part of at least one cloud mass. More preferably, the second region Ar2 includes at least the boundary between the extracted cloud mass and the sky. More preferably, the second region Ar2 includes all the boundaries of the extracted cloud mass with the sky. In the present embodiment, as shown in FIG. 14, the second region Ar21 is set for the cloud mass m1, the second region Ar22 is set for the cloud mass m2, and the second region Ar23 is set for the cloud mass m3. Is set, the second region Ar24 is set for the cloud mass m4, and the second region Ar25 is set for the cloud mass m5. These examples are set to include all the boundaries of one cloud mass with the sky. For example, the second region Ar21 includes all the boundaries of the cloud mass m1 with the sky.

<Second cloud position calculation unit 15>
The second cloud position calculation unit 15 shown in FIG. 1 is a second cloud position calculation unit 15 based on a comparison of images corresponding to the second regions Ar21 to Ar25 in the sky images acquired from at least two cameras 10 among the plurality of sky images. The position information of the cloud mass in the areas Ar21 to Ar25 is calculated. As an example, the second region Ar21 shown in FIG. 14 is taken. First, the two cameras 10 (C1, C2) closest to the second region Ar21 are selected, and the aerial images G1 and G2 obtained from those two cameras 10 are acquired. Using these two sky images G1 and G2, cloud position information [latitude / longitude (cloud distribution matrix A), cloud altitude matrix] is calculated for the second region Ar21. The method of calculating the cloud position information is the same as that of the first cloud position calculation unit 12. Similarly, the cloud position information is calculated for the second regions Ar22, Ar23, Ar24, and Ar25. Regarding this embodiment, in the processing by the first cloud position calculation unit 12, the altitude was searched in the range of 0 to 4000 m, but the second cloud position calculation unit 15 searched in the range of ± 500 m centered on 1800 m. Since the altitude of the cloud is estimated in advance in this way, it is possible to narrow the search range and reduce the calculation cost.

When the calculation of cloud position information is completed for all the second regions, each calculation result is overwritten in the first region Ar1. The results are shown in FIG. As shown in FIG. 15, the altitude of the cloud mass m1 was 1948 m, the altitude of the cloud mass m2 was 1934 m, the altitude of the cloud mass m3 was 1905 m, the altitude of the cloud mass m4 was 1965 m, and the altitude of the cloud mass m5 was 1750 m. .. It is possible to calculate different altitudes for each cloud mass.

[Method]
The information processing method executed by the system 1 will be described with reference to FIG.

First, in step ST100, the image acquisition unit 11 acquires aerial images G1, G2, and G3 including at least the sky taken by a plurality of cameras 10 arranged at different positions from each other. In the next step ST101, the first cloud position calculation unit 12 determines the clouds in the first region Ar1 based on the comparison of the sky images acquired from at least two cameras 10 among the plurality of sky images G1, G2, and G3. Calculate the position information of. A detailed explanation will be described later with reference to FIG. In the next step ST102, the extraction unit 13 extracts cloud masses m1 to m5 from the cloud position information in the first region Ar1. In the next step ST103, the second area setting unit 14 sets the second area Ar2 (Ar21, Ar22, Ar23, Ar24, Ar25) including the extracted cloud masses m1 to m5 and narrower than the first area Ar1. In the next step ST104, the second cloud position calculation unit 15 determines the second region Ar2 (Ar21, Ar22, Ar23, in the aerial image acquired from at least two cameras 10 among the plurality of aerial images G1, G2, G3. Based on the comparison of the images corresponding to Ar24, Ar25), the position information of the cloud mass in the second region Ar2 (Ar21, Ar22, Ar23, Ar24, Ar25) is calculated. The processes of steps ST103 and ST104 are executed for each cloud mass.

Next, a method of calculating the cloud position information in the first region Ar1 will be described with reference to FIG. In step ST200, the common area setting unit 12a sets the common area Ar1 which is also the first area Ar1 based on the camera installation points C1 to C4 as shown in FIG. Next, in step ST201, the subregion setting unit 12b generates all the sets of the two cameras 10 from all the cameras 10 (C1 to C4), and sets the subregion Ar10 for each set. In the next step 202, the subregion cloud position calculation unit 12c acquired the cloud position information in the set subregion Ar10 (third region Ar13, fourth region Ar14) from the corresponding two cameras 10. Calculated based on the comparison of aerial images. Here, as shown in FIG. 12, the first region Ar1 includes the third region Ar13 and the fourth region Ar14. The subregion cloud position calculation unit 12c calculates the cloud position information in the third region Ar13 based on the first set of sky images (G1, G2) acquired from at least two cameras 10. The subregion cloud position calculation unit 12c is based on the sky images (G2, G3) of the second set, which is a combination different from the first set, acquired from at least two cameras 10, and the cloud position calculation unit 12c in the fourth region Ar14. Calculate location information. In the next step ST203, the synthesis unit 12d synthesizes the calculated cloud position information of each subregion Ar10 (including the third region Ar13 and the fourth region Ar14) into the common region Ar1 and in the first region Ar1. Calculate cloud position information. Here, the cloud position information in the third region Ar13 and the cloud position information in the fourth region Ar14 are combined to calculate the cloud position information in the first region Ar1.

As described above, the cloud position calculation system 1 of the present embodiment includes an image acquisition unit 11 that acquires at least aerial images G1 to G3 including the sky taken by a plurality of cameras 10 arranged at different positions. The first cloud position calculation unit 12 that calculates the cloud position information in the first area Ar1 based on the comparison of the sky images acquired from at least two cameras 10 among the sky images G1 to G3, and the first area. An extraction unit 13 that extracts cloud masses m1 to m5 from cloud position information in Ar1, and a second area setting unit 14 that includes the extracted cloud masses m1 to m5 and sets a second region Ar2 that is narrower than the first region Ar1. And, among the plurality of sky images G1 to G3, the position information of the cloud mass in the second region Ar2 is calculated based on the comparison of the images corresponding to the second region Ar2 in the sky images acquired from at least two cameras 10. A second cloud position calculation unit 15 is provided.

The cloud position calculation method of the present embodiment acquires aerial images G1 to G3 including at least the sky taken by a plurality of cameras 10 arranged at different positions (ST100), and a plurality of aerial images G1 to G3. Of these, the cloud position information in the first region Ar1 is calculated (ST101) based on the comparison of the sky images acquired from at least two cameras 10, and the cloud mass is calculated from the cloud position information in the first region Ar1. Extracting m1 to m5 (ST102), setting a second area Ar2 including the extracted cloud masses m1 to m5 and narrower than the first area Ar1 (ST103), and setting a plurality of aerial images G1 to G3. Among them, the calculation of the position information of the cloud mass in the second region Ar2 based on the comparison of the images corresponding to the second region Ar2 in the sky images acquired from at least two cameras 10 (ST104) is included.

In this way, the cloud position information in the relatively wide first region Ar1 is calculated based on the comparison of the aerial images, and the second region narrower than the first region Ar1 including the cloud mass extracted from the cloud position information. Since the position information of the cloud masses m1 to m5 is calculated by comparing the aerial images with Ar2 as the target, the position information for each cloud mass is compared with the case where the comparison is performed focusing on a part of the cloud masses of the aerial image from the beginning. Can be calculated accurately. Nevertheless, first, since the position information of the clouds in the relatively wide first region Ar1 is calculated, the consistency as a whole can be ensured, and then the position information of the local cloud mass in the second region can be estimated with an error. It will be possible to reduce it.

As in the system of the present embodiment, the first region Ar1 includes the third region Ar13 and the fourth region Ar14, and the first cloud position calculation unit 12 is a set of the first set acquired from at least two cameras 10. Based on the aerial images (G1, G2), the position information of the clouds in the third region Ar13 is calculated, and the second set of aerial images (a combination different from the first set) acquired from at least two cameras 10 (a second set of aerial images (G1, G2). Based on G2 and G3), the cloud position information in the fourth region Ar14 is calculated, and the cloud position information in the third region Ar13 and the cloud position information in the fourth region Ar14 are combined to form the first region. It is preferable to calculate the position information of the cloud in Ar1.

According to this configuration, the position information of the clouds in the first region Ar1 is calculated by synthesizing the position information of the clouds in a plurality of regions obtained from a plurality of sets of aerial images, so that the position information of a wide range of clouds can be calculated accurately. It becomes.

As in the system of the present embodiment, the first cloud position calculation unit 12 calculates the reliability ZNCC ₃ of the third region Ar13 together with the cloud position information in the third region Ar13, and the cloud position information in the fourth region Ar14. _{When the reliability ZNCC 4} of the 4th region Ar14 is calculated together with, and the 3rd region Ar13 and the 4th region Ar14 overlap, the cloud position information at the overlapping location is obtained from the corresponding cloud of the 3rd region Ar13 and the 4th region Ar14. It is preferable to calculate based on the position information of the above and the _{weights of the respective reliability ZNCC 3} and ZNCC _4.

According to this configuration, the position information of the clouds at the overlapping points _{is calculated with the reliability ZNCC 3} and ZNCC ₄ of each region as weights, so that the position information of a wide range of clouds can be acquired while improving the accuracy of the cloud position information. It will be possible.

As in the system of the present embodiment, it is preferable that the second cloud position calculation unit 15 uses sky images acquired from at least two cameras 10 near the second area Ar2.

As described above, if the camera 10 is close to the second region Ar2, the height component of the clouds in the second region Ar2 reflected in the aerial image can be reduced, and the accuracy can be improved.

Like the system of the present embodiment, it is preferable that the second region Ar2 includes at least the boundary between the extracted cloud masses m1 to m5 and the sky.

As described above, since the second region Ar2 includes at least the boundary between the extracted cloud masses m1 to m5 and the sky, features such as the shape and arrangement of the cloud masses can be taken into consideration, and the accuracy can be improved. ..

As in the system of the present embodiment, it is preferable that the cloud position information includes at least one of cloud altitude information and latitude / longitude information. This is a preferred embodiment.

As in the system of the present embodiment, it is preferable to calculate the position information of the clouds by pattern matching between pixels indicating clouds in the sky image. This is a preferred embodiment.

Like the system of the present embodiment, the first region Ar1 or the second region Ar2 is meshed along the latitude and longitude, and it is preferable that the cloud position information is represented for each cell of the mesh. This is a preferred embodiment.

It is preferable to further include a plurality of cameras arranged at different positions as in the system of the present embodiment. This is a preferred embodiment.

The program according to this embodiment is a program that causes a computer to execute the above method. Further, the temporary recording medium readable by the computer according to the present embodiment stores the above program.

Although the embodiments of the present disclosure have been described above based on the drawings, it should be considered that the specific configuration is not limited to these embodiments. The scope of the present disclosure is set forth not only by the description of the embodiment described above but also by the scope of claims, and further includes all modifications within the meaning and scope equivalent to the scope of claims.

For example, the execution order of each process such as operation, procedure, step, and step in the device, system, program, and method shown in the claims, specification, and drawings may be the output of the previous process. It can be realized in any order unless it is used in processing. Even if the scope of claims, the specification, and the flow in the drawings are explained using "first", "next", etc. for convenience, it does not mean that it is essential to execute in this order. ..

Each part 12 to 17 shown in FIG. 1 is realized by executing a predetermined program by 1 or a processor, but each part may be configured by a dedicated memory or a dedicated circuit.

In the system 1 of the above embodiment, each part 11 to 19 is mounted on the processor 1b of one computer, but each part 11 to 19 may be distributed and mounted on a plurality of computers or the cloud. That is, the above method may be executed by one or more processors.

It is possible to adopt the structure adopted in each of the above embodiments in any other embodiment. In FIG. 1, for convenience of explanation, each part 11 to 19 is mounted, but some of them can be omitted arbitrarily. For example, an embodiment in which each part 11 to 17 is mounted can be mentioned.

The specific configuration of each part is not limited to the above-described embodiment, and various modifications can be made without departing from the gist of the present disclosure.

1 Cloud position calculation system 11 Image acquisition unit 12 1st cloud position calculation unit 13 Extraction unit 14 2nd area setting unit 15 2nd cloud position calculation unit

Claims

An image acquisition unit that acquires at least an aerial image including the sky taken by multiple cameras arranged at different positions.
A first cloud position calculation unit that calculates cloud position information in a first area based on a comparison of the sky images acquired from at least two cameras among the plurality of sky images.
An extraction unit that extracts cloud masses from the position information of clouds in the first area,
A second area setting unit that includes the extracted cloud mass and sets a second area narrower than the first area,
The second cloud that calculates the position information of the cloud mass in the second region based on the comparison of the images corresponding to the second region in the sky image acquired from at least two cameras among the plurality of sky images. Position calculation unit and
A cloud position calculation system equipped with.
The system according to claim 1.
The first region includes the third region and the fourth region, and includes the third region and the fourth region.
The first cloud position calculation unit calculates cloud position information in the third region based on the first set of aerial images acquired from at least two cameras.
Based on the sky images of the second set, which is a combination different from the first set, acquired from at least two cameras, the position information of the clouds in the fourth area is calculated.
A cloud position calculation system that calculates the position information of clouds in the first area by synthesizing the position information of clouds in the third area and the position information of clouds in the fourth area.
The system according to claim 2.
The first cloud position calculation unit calculates the reliability of the third region together with the cloud position information in the third region, and calculates the reliability of the fourth region together with the cloud position information in the fourth region. ,
When the third region and the fourth region overlap, the cloud position information at the overlapping location is the corresponding cloud position information of the third region and the fourth region, and the weight which is the reliability of each of them. Cloud position calculation system that calculates based on.
The system according to any one of claims 1 to 3.
The second cloud position calculation unit is a cloud position calculation system that uses sky images acquired from at least two cameras near the second area.
The system according to any one of claims 1 to 4.
The second area is a cloud position calculation system including at least the boundary of the extracted cloud mass with the sky.
The system according to any one of claims 1 to 5.
The cloud position information is a cloud position calculation system including at least one of the altitude information and the latitude / longitude information of the cloud.
The system according to any one of claims 1 to 6.
A cloud position calculation system that calculates the position information of the clouds by pattern matching between pixels showing clouds in the sky image.
The system according to any one of claims 1 to 7.
A cloud position calculation system in which the first region or the second region is meshed along latitude and longitude, and the cloud position information is expressed for each cell of the mesh.
The system according to any one of claims 1 to 8.
A cloud position calculation system with multiple cameras located at different positions.
Acquiring at least aerial images including the sky taken by multiple cameras located at different positions,
To calculate the position information of clouds in the first region based on the contrast of the sky images acquired from at least two cameras among the plurality of sky images.
Extracting a cloud mass from the cloud position information in the first area and
To set a second area that includes the extracted cloud mass and is narrower than the first area.
To calculate the position information of the cloud mass in the second region based on the comparison of the images corresponding to the second region in the sky image acquired from at least two cameras among the plurality of aerial images.
Cloud position calculation method including.
A program that causes one or more processors to execute the method according to claim 10.