All-day fine cloud detection device and method

Abstract

The application discloses a kind of all-day fine cloud detection device and method, suitable for the cloud amount monitoring of various automatic astronomical observation or space target observation site.The device includes heating device, transparent protective cover, wide-angle optical lens, light shielding component, CCD visible light detector, embedded control board.In the method, after the calibration of the pointing model of detection device using star matching algorithm, use GPS information to judge whether it is daytime or night;According to the judgment result, different algorithms are selected to complete all-day cloud detection;Finally, according to the pointing provided by other equipment, the fine cloud amount corresponding to the pointing is given.Through the above device and method, the application can not only complete all-day cloud detection, but also detect the cloud amount of a certain observation area at a certain pointing at low cost;At the same time, according to the needs of astronomical observation and space target observation, the cloud amount is divided using star magnitude at night, and the cloud amount division is more fine.

Description

Technical Field

[0001] This invention relates to the field of meteorological observation, and in particular to a fine-grained all-day cloud detection device and method. Background Technology

[0002] Clouds cover more than 50% of the Earth's surface and are a crucial meteorological element. Especially for ground-based astronomical observations or space target detection, the presence of clouds can severely impair their effectiveness. Efficient and precise all-sky cloud detection technology is an important means to improve the effectiveness of ground-based astronomical observations or space target detection, and it is also a necessary condition for the automated operation of ground-based observation systems.

[0003] Current methods for detecting cloud cover on ground can be categorized into visual inspection, inversion methods, and remote sensing methods. These methods have the following drawbacks:

[0004] 1) High detection cost: Current cloud cover detection either requires high manpower costs or expensive measurement equipment such as lasers or multispectral cameras;

[0005] 2) Limited detection precision: Current cloud cover detection equipment generally only provides the total cloud cover of the detected area, and cannot accurately provide the cloud cover of a specific area pointing downwards, which is not suitable for this purpose;

[0006] 3) Limited equipment integration: Current cloud data detection equipment generally only provides data acquisition functions and requires a separate PC or processing system for processing, resulting in limited real-time performance and integration. Summary of the Invention

[0007] The purpose of this invention is to provide a fine-grained cloud detection device and method for all-day observation, which can solve the problems of high cloud detection cost, imprecise detection results, and limited integration of detection equipment in astronomical observation or space target detection.

[0008] To address the aforementioned technical problems, one technical solution of the present invention is to provide an all-day cloud detection device, comprising:

[0009] Heating device;

[0010] A transparent protective cover, filled with dry air, works in conjunction with the heating device to provide a dry and warm environment inside the cover, preventing condensation inside the cover;

[0011] A wide-angle optical lens, housed within a transparent protective housing, provides a 180° field of view.

[0012] Light-shielding components are used to block stray light from the ground and the surrounding environment;

[0013] CCD visible light detectors are used to acquire visible light images;

[0014] The embedded control board provides control signals to the CCD visible light detector and heating device connected to it, and processes the cloud map in real time.

[0015] Another technical solution of the present invention is to provide an all-day cloud detection method, which is implemented based on an all-day cloud detection device, and the method includes the following steps:

[0016] S1. Install the cloud detection device with its line of sight aligned with the zenith direction;

[0017] S2. On a clear night, the all-sky cloud detection device is used to capture star images. The embedded control board processes the star images and calibrates the device based on star image matching to calculate the pointing model.

[0018] S3. Control the all-day cloud detection device to capture cloud images. The embedded control board obtains local geographical location information and time information through GPS and calculates the local sunrise and sunset times by combining the solar orbit parameters.

[0019] S4. If the shooting time is during the day, first detect the cloud area by using texture and grayscale detection methods, and then combine the sun's orbit parameters to eliminate the influence of the sun on cloud detection in the field of view, and obtain the cloud distribution map for the whole day.

[0020] S5. If the shooting time is at night, firstly, the cloud layer and natural celestial body in the image are distinguished by the filtering algorithm. The cloud layer area is detected by the texture and grayscale detection method. Combined with the lunar orbit, the influence of the moon on the cloud detection is eliminated, and the cloud layer distribution map of the whole sky is obtained.

[0021] S6. Based on the all-day cloud distribution map obtained in step S4 or S5, and combined with the pointing model in step S2, calculate the cloud cover value for any pointing direction to complete the cloud cover detection.

[0022] Before use, the all-day fine cloud detection device of this invention requires calibration of its pointing model using a star-matching algorithm on a clear night. After calibration, it autonomously determines whether the current time is day or night using GPS information. Based on the determination result, different algorithms are selected to complete all-day cloud detection. Finally, based on the pointing information provided by other devices, the device provides the fine cloud cover corresponding to this pointing direction. The device and method of this invention are applicable to cloud cover monitoring at various automated astronomical observation or space target observation stations.

[0023] The beneficial effects of this invention are:

[0024] 1) Low testing cost: This invention can achieve fully autonomous calibration and testing. The key components are a visible light detector and an embedded control board. The labor and equipment costs are lower than those of current solutions.

[0025] 2) Limited detection precision: This invention can detect cloud cover not only throughout the day, but also cloud cover in a specific observation area pointing downwards; at the same time, in response to the needs of astronomical observation and space target observation, it uses star magnitude to divide cloud cover at night, making the cloud cover division more precise.

[0026] 3) High equipment integration: This invention integrates optical lenses, detectors, GPS and embedded controllers into a whole, with high equipment integration. It can directly output detection results according to user needs without the need for additional processing equipment. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the all-day fine cloud detection device described in this invention;

[0028] Figure 2 This is a schematic diagram of the overall process of the all-day fine cloud detection method described in this invention;

[0029] Figure 3 This is a schematic diagram illustrating the model building process in the method described in this invention;

[0030] Figure 4 This is a schematic diagram of the daytime cloud detection process in the method described in this invention;

[0031] Figure 5 This is a schematic diagram of the nighttime cloud detection process in the method described in this invention;

[0032] Figure 6 This is a schematic diagram of the cloud cover calculation process in the method described in this invention. Detailed Implementation

[0033] like Figure 1 As shown, the all-day fine cloud detection device provided by the present invention includes a transparent protective cover 1, a wide-angle optical lens 2, a detachable level 3, a light-shielding component 4, a heating device 5, a CCD visible light detector 6, a GPS antenna 7, a network antenna 8, a sealed housing 9, an embedded control board 10, a power plug 11 and a power cord, and a heat sink 12.

[0034] For example, the transparent protective cover 1 can be a glass protective cover, with the wide-angle optical lens 2 located inside it; the light-shielding assembly 4 is located below the glass protective cover, and has a laterally extending extension around the glass protective cover, with the outer edge of the extension facing upwards; a detachable level 3 can be placed at the extension. The sealed housing 9 is located below the light-shielding assembly 4, and the heating device 5 and CCD visible light detector 6 are located inside the sealed housing 9 and are respectively connected to the embedded control board 10; the heating device 5 can be an annular heating band, which in this example is located at the top inside the sealed housing 9, close to the glass protective cover. In this example, the embedded control board 10 encloses the bottom of the sealed housing 9; the heat sink 12 can be a semiconductor heat sink, which in this example is located below the embedded control board 10; the GPS antenna 7, network antenna 8, power plug 11, and power cord are outside the sealed housing 9 and are each connected to the embedded control board 10.

[0035] The glass protective cover is filled with dry air, which, together with the annular heating belt, provides a dry and warm environment inside the cover, preventing condensation. The wide-angle optical lens 2 provides the cloud detection device with a 180° field of view. The detachable level 3 is used to ensure that the optical axis is aligned with the zenith direction when installing the cloud detection device; the level 3 is removed after the device is installed.

[0036] The light-shielding component 4 is used to shield stray light from the ground and surrounding environment. The CCD visible light detector 6 is used to acquire visible light images. The GPS antenna 7 provides a time reference for the embedded control board 10 and the CCD visible light detector 6. The network antenna 8 is used to receive external control signals or remotely transmit cloud detection results.

[0037] The embedded control board 10 provides control signals to the CCD visible light detector 6, the annular heating strip 5, and the semiconductor heat sink 12, while simultaneously processing the cloud image in real time. In this example, the embedded control board 10 is equipped with a Linux kernel operating system. The power plug 11 and power cord supply power to the device. The semiconductor heat sink 12 cools the electronic components within the sealed housing 9, ensuring that the temperature inside the housing 9 remains within a suitable range.

[0038] like Figure 2 As shown, the all-day fine-grained cloud detection method provided by this invention includes the following steps:

[0039] S1. Install the cloud detection device with its line of sight aligned with the zenith direction;

[0040] S2. On a clear night, the cloud detection device is used to capture star images. The embedded control board processes the star images and calibrates the device based on star image matching to calculate the pointing model.

[0041] S3. Control the cloud detection device to capture cloud images. The embedded control board obtains local geographical location information and time information through GPS and calculates the local sunrise and sunset times by combining the solar orbit parameters.

[0042] S4. If the shooting time is during the day, first detect the cloud area by using texture and grayscale detection methods, and then combine the sun's orbit parameters to eliminate the influence of the sun on cloud detection in the field of view, and obtain the cloud distribution map for the whole day.

[0043] S5. If the shooting time is at night, firstly, the cloud layer and natural celestial body in the image are distinguished by the filtering algorithm. The cloud layer area is detected by the texture and grayscale detection method. Combined with the lunar orbit, the influence of the moon on the cloud detection is eliminated, and the cloud layer distribution map of the whole sky is obtained.

[0044] S6. Based on the cloud distribution map obtained in step S4 or S5, and combined with the pointing model in step S2, the cloud amount value for a certain direction can be calculated to complete the fine detection of cloud amount.

[0045] Step S2 only needs to be executed once after the all-day fine cloud detection device is deployed. It is used to establish the transformation relationship from image plane coordinates to the horizon coordinate system through the star matching algorithm, that is, the pointing model.

[0046] like Figure 3 As shown, in step S2 of the present invention, the step of pointing to the model calibration is as follows:

[0047] S2-1. The embedded control board controls the visible light CCD detector to acquire starry sky images. The control board reads the starry sky images, acquisition time, device location and other data into memory.

[0048] S2-2, The embedded control board performs flat-field dark-field correction on the starry sky image, detects star points in the image, and calculates the centroid coordinates (x, y) of each star point in the image plane coordinate system. i ,y i ), i∈[1,N];

[0049] S2-3. The embedded control board, based on the star map acquisition time and device location information, uses a coordinate transformation algorithm to convert the geographical latitude and longitude of the device at the time of image acquisition into the right ascension and declination in the J2000 coordinate system, and uses this right ascension and declination as the right ascension and declination of the local zenith at the acquisition time.

[0050] S2-4. The embedded control board reads star data within a 10° range in the star catalog based on the right ascension and declination obtained in step S2-3, using this as the center.

[0051] S2-5. The embedded control board selects an area of ​​approximately 10°*10° from the starry sky image and selects the centroid coordinates of the star points within the area.

[0052] S2-6. Using the triangle matching algorithm, match the centroid coordinates of star points in the image with the celestial coordinates of stars in the star catalog. The celestial coordinates of the stars after matching are represented as (α... i ,δ i ), i∈[1,N];

[0053] S2-7. Considering factors such as precession, nutation, proper motion of stars, atmospheric aberration, phototropic aberration, and the location of the instrument deployment, the celestial coordinates of the stars are converted into the azimuth and altitude angles below the horizon, expressed as (a i ,h i ), i∈[1,N];

[0054] S2-8. Transform the star-centric horizontal coordinates into ideal coordinates on the tangent plane. The transformation method is as follows:

[0055]

[0056] Where (a0,h0) is the azimuth and elevation angle corresponding to the center of the starry sky image.

[0057] S2-9. Establish the transformation relationship between the centroid coordinates of star points in the image and the ideal coordinates of stars in step S2-8 using a polynomial model, and calculate the parameter values ​​in the polynomial model using the least squares method:

[0058]

[0059]

[0060] Where, x i and y i Let ζ be the centroid coordinates of the star point sequence in the image. i and η i Let a1~j1 and a2~j2 be the ideal coordinates of the star sequence, and let a1~j1 and a2~j2 be the polynomial coefficients in the two formulas, respectively.

[0061] S2-10. To mitigate the impact of field-of-view distortion on star matching, the selected range is expanded in 10° increments. Steps S2-4 to S2-9 are repeated until the entire field of view is reached, resulting in a full-field-of-view pointing model.

[0062] like Figure 4 As shown, in step S4 of the present invention, the daytime cloud detection step is as follows:

[0063] S4-1. First, divide the original cloud image into small regions of M0*N0, each of which occupies an area of ​​m0*n0 pixels.

[0064] S4-2. Use the LBP algorithm to calculate the LBP texture features of each small block region, and perform statistics on the LBP texture feature values ​​within the region.

[0065] S4-3. Classify the LBP texture feature statistics obtained in step S4-2, and label different image regions as thick cloud areas, cloudy areas, cloudless areas, and uncertain areas.

[0066] S4-4. Calculate the average gray value in different blocks, divide the uncertain areas into thick cloud areas or cloudy areas according to the similarity of gray values, and divide the remaining blocks into cloudless areas. Finally, divide the image area into thick cloud areas, cloudy areas and cloudless areas.

[0067] S4-5. Calculate the coordinates of the sun in the image at the time of shooting, taking into account the image's shooting time, the device's geographical location, the device's pointing model, and the sun's orbit.

[0068] S4-6. Remove the image region with the sun coordinates and its neighborhood; combine the thick cloud region and the clouded region, and use the Ostu adaptive threshold segmentation algorithm to calculate the thick cloud segmentation threshold T1; combine the clouded region and the cloudless region, and use the Ostu adaptive threshold segmentation algorithm to calculate the thin cloud segmentation threshold T2.

[0069] S4-7. Use the two thresholds from step S4-6 to segment the entire image, and finally divide the image into three regions: thick cloud region, cloudy region and cloudless region.

[0070] S4-8. Using Hough transform, detect circles near the sun coordinates in the classification image obtained in step S4-7. If a circular area can be detected, it means that the sun is not obscured by clouds, and the circular area is marked as a cloudless area; if a circular area cannot be detected, it means that the sun is obscured by clouds, and the original markings in the image are retained; finally, a cloud distribution map of the entire sky is obtained.

[0071] like Figure 5 As shown, in step S5 of the present invention, the nighttime cloud detection step is as follows:

[0072] S5-1. Let the original image be I0. Perform median filtering on the original image to obtain the filtered image I1. Divide the original image and the filtered image pixel by pixel to obtain the third image I2. Since clouds in the image exhibit low-frequency features and stars exhibit high-frequency features, image I1 retains most of the cloud features and is called a cloud image; image I2 retains most of the star features and is called a star image.

[0073] S5-2. Divide the cloud map into small regions of M0*N0, each of which occupies an area of ​​m0*n0 pixels;

[0074] S5-3. Use the LBP algorithm to calculate the LBP texture features of each small block region, and perform statistics on the LBP texture feature values ​​within the region.

[0075] S5-4. Classify the LBP texture feature statistics obtained in step S5-3, and label different image regions as clouded areas, cloudless areas, and uncertain areas.

[0076] S5-5. Calculate the average gray value in different blocks, divide the uncertain area into clouded and cloudless areas according to the similarity of gray values, and finally divide the image area into clouded and cloudless areas.

[0077] S5-6. Calculate the coordinates of the moon in the image at the time of capture, taking into account the image's capture time, the device's geographical location, the device's pointing model, and the moon's orbit.

[0078] S5-7. Remove the image region containing the lunar coordinates and its neighborhood; calculate the segmentation threshold T for the remaining region using the Ostu adaptive threshold segmentation algorithm;

[0079] S5-8. Use the threshold from step S5-7 to segment the entire image, ultimately dividing the image into two regions: a cloudy region and a cloudless region.

[0080] S5-9. Detect the connected regions near the lunar coordinates in step S5-8, and calculate whether the connected regions conform to the theoretical field of view of the moon. If the connected regions conform to the theoretical field of view of the moon, it proves that the moon is not obscured by clouds, and this connected region is marked as a cloudless area; if the connected regions are far beyond the lunar field of view, it proves that the moon is obscured by clouds, and the original marking is retained.

[0081] S5-10. Using the cloud area as a mask, detect the star points in the cloud area of ​​the star map, and count the density distribution of different star magnitudes in the cloud area. Divide the cloud thickness into 6 levels according to the magnitude of stars 6-10, with the higher the level, the thicker the cloud. If the stars of magnitudes 6-10 in a certain sub-region cannot reach the theoretical density, it is defined as level 6; if only the stars of magnitude 6 can reach the theoretical density, it is defined as level 5, and so on.

[0082] S5-11. Define the cloudless area as level 0. All pixels in the final cloud map are marked as 7 levels from 0 to 6, where 0 represents no cloud and 7 represents full cloud. Use this marked map as the cloud distribution map.

[0083] Step S6 can combine the optical axis direction and field of view of the observation equipment to project the observation range of the observation equipment onto the cloud distribution map, thereby calculating the cloud cover value of the observation area of ​​the observation equipment.

[0084] like Figure 6 As shown, in step S6 of the present invention, the cloud cover calculation step for the pointed area is as follows:

[0085] S6-1. Receive pointing and field-of-view information from the astronomical telescope via a network antenna;

[0086] S6-2. Calculate the azimuth and elevation angles corresponding to the edges of the field of view;

[0087] S6-3. Using the pointing model, the azimuth and elevation angle sequences from step S6-2 are converted into image plane coordinate sequences to determine the area to be identified;

[0088] S6-4. Calculate the cloud cover within the area specified in step S6-3;

[0089] S6-5. Cloud cover data is fed back to the observation telescope via the network.

[0090] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for all-day cloud detection, characterized in that, This is implemented based on an all-day cloud detection device, which includes: Heating device; A transparent protective cover, filled with dry air, works in conjunction with the heating device to provide a dry and warm environment inside the cover, preventing condensation inside the cover; A wide-angle optical lens, housed within a transparent protective housing, provides a 180° field of view. Light-shielding components are used to block stray light from the ground and the surrounding environment; CCD visible light detectors are used to acquire visible light images; The embedded control board provides control signals to the CCD visible light detector and heating device connected to it, and processes cloud images in real time. The method includes the following steps: S1. Align the line of sight of the all-sky cloud detection device with the zenith direction and install it. S2. On a clear night, the all-sky cloud detection device is used to capture star images. The embedded control board processes the star images and calibrates the device based on star image matching to calculate the pointing model. S3. Control the all-day cloud detection device to capture cloud images. The embedded control board obtains local geographical location information and time information through GPS and calculates the local sunrise and sunset times by combining the solar orbit parameters. S4. If the shooting time is during the day, first detect the cloud area by using texture and grayscale detection methods, and then combine the sun's orbit parameters to eliminate the influence of the sun on cloud detection in the field of view, and obtain the cloud distribution map for the whole day. S5. If the shooting time is at night, firstly, the cloud layer and natural celestial body in the image are distinguished by the filtering algorithm. The cloud layer area is detected by the texture and grayscale detection method. Combined with the lunar orbit, the influence of the moon on the cloud detection is eliminated, and the cloud layer distribution map of the whole sky is obtained. S6. Based on the all-day cloud distribution map obtained in step S4 or S5, and combined with the pointing model in step S2, calculate the cloud cover value for any pointing direction to complete the cloud cover detection. Step S5 further includes: S5-1. Let the original image be... The original image is subjected to median filtering to obtain the filtered image. The original image is divided pixel by pixel by the filtered image to obtain the third image. Because clouds in the image exhibit low-frequency characteristics, while stars exhibit high-frequency characteristics, therefore the image... The image retains most of the cloud features and is called a cloud image; It retains most of the star features and is called a star map; S5-2, Divide the cloud map into... Small regions, each occupying an area of Pixel; S5-3. Use the LBP algorithm to calculate the LBP texture features of each small block region, and perform statistics on the LBP texture feature values ​​within the region. S5-4. Classify the LBP texture feature statistics obtained in step S5-3, and label different image regions as clouded areas, cloudless areas, and uncertain areas. S5-5. Calculate the average gray value in different blocks, divide the uncertain area into clouded and cloudless areas according to the similarity of gray values, and finally divide the image area into clouded and cloudless areas. S5-6. Calculate the coordinates of the moon in the image at the time of capture, taking into account the image's capture time, the device's geographical location, the device's pointing model, and the moon's orbit. S5-7. Remove the image region containing the lunar coordinates and its neighborhood; calculate the segmentation threshold T for the remaining region using the Ostu adaptive threshold segmentation algorithm; S5-8. Use the threshold from step S5-7 to segment the entire image, ultimately dividing the image into two regions: a cloudy region and a cloudless region. S5-9. Detect the connected regions near the lunar coordinates in step S5-8, and calculate whether the connected regions conform to the theoretical field of view of the moon. If the connected regions conform to the theoretical field of view of the moon, it proves that the moon is not obscured by clouds, and this connected region is marked as a cloudless area; if the connected regions are far beyond the lunar field of view, it proves that the moon is obscured by clouds, and the original marking is retained. S5-10. Using the cloud area as a mask, detect the star points in the cloud area of ​​the star map, and count the density distribution of different star magnitudes in the cloud area. Divide the cloud thickness into 6 levels according to the stars of magnitude 6-10, with the higher the level, the thicker the cloud. For the sub-region where the theoretical density cannot be reached by stars of magnitude 6-10, it is defined as level 6; for the sub-region where only stars of magnitude 6 can reach the theoretical density, it is defined as level 5, and so on. S5-11. Define the cloudless area as level 0. In the final cloud map, all pixels are marked as 7 levels from 0 to 6, where 0 represents no cloud and 7 represents full cloud, thus obtaining the cloud distribution map.

2. The all-day cloud detection method as described in claim 1, characterized in that, Step S2 is executed once after the all-sky cloud detection device is deployed. It is used to establish the transformation relationship from image plane coordinates to the horizon coordinate system, i.e., the pointing model, through the star matching algorithm.

3. The all-day cloud detection method as described in claim 1, characterized in that, Step S4 further includes: S4-1, First, divide the original cloud map into... Small regions, each occupying an area of Pixel; S4-2. Use the LBP algorithm to calculate the LBP texture features of each small block region, and perform statistics on the LBP texture feature values ​​within the region. S4-3. Classify the LBP texture feature statistics obtained in step S4-2, and label different image regions as thick cloud areas, cloudy areas, cloudless areas, and uncertain areas. S4-4. Calculate the average gray value in different blocks, divide the uncertain areas into thick cloud areas or cloudy areas according to the similarity of gray values, and divide the remaining blocks into cloudless areas. Finally, divide the image area into thick cloud areas, cloudy areas and cloudless areas. S4-5. Calculate the coordinates of the sun in the image at the time of shooting, taking into account the image's shooting time, the device's geographical location, the device's pointing model, and the sun's orbit. S4-6. Remove the image region containing the sun coordinates and its neighborhood; combine the thick cloud region and the cloudy region, and use the Ostu adaptive thresholding segmentation algorithm to calculate the thick cloud segmentation threshold. Combining cloud-covered and cloudless areas, the Ostu adaptive threshold segmentation algorithm is used to calculate the thin cloud segmentation threshold. ; S4-7. Use the two thresholds from step S4-6 to segment the entire image, and finally divide the image into three regions: thick cloud region, cloudy region and cloudless region. S4-8. Using Hough transform, detect circles near the sun coordinates in the classification image obtained in step S4-7. If a circular area can be detected, it means that the sun is not obscured by clouds, and the circular area is marked as a cloudless area; if a circular area cannot be detected, it means that the sun is obscured by clouds, and the original markings in the image are retained; finally, a cloud distribution map of the entire sky is obtained.

4. The all-day cloud detection method as described in claim 1, characterized in that, In step S6, the observation range of the observation device is projected onto the cloud distribution map by combining the optical axis direction and field of view of the observation device, and the cloud cover value of the observation area of ​​the observation device is calculated.

5. The all-day cloud detection method as described in claim 1, characterized in that, The transparent protective cover is a glass protective cover; The heating device is a ring-shaped heating belt.

6. The all-day cloud detection method as described in claim 1, characterized in that, The all-day cloud detection device also includes: A sealed housing for mounting a CCD visible light detector, heating device, and embedded control board; The semiconductor heat sink is used to cool the electronic components inside the sealed housing, ensuring that the temperature inside the housing is within a suitable range; the embedded control board is also used to provide control signals to the semiconductor heat sink.

7. The all-day cloud detection method as described in claim 1, characterized in that, The all-day cloud detection device also includes: A detachable level is used to ensure that the optical axis is aligned with the zenith direction when installing the all-sky cloud detection device. The level is removed after the device is installed.

8. The all-day cloud detection method as described in claim 1, characterized in that, The all-day cloud detection device also includes: A network antenna, which is connected to the embedded control board, is used to receive external control signals or remotely send cloud detection results; The GPS antenna is connected to the embedded control board and is used to provide a time reference for the embedded control board and the CCD visible light detector.

Images