Meteorological measurement device

The weather measuring device addresses maintenance and clogging issues by using machine learning and image detection with an infrared camera to estimate falling particles, offering accurate and reliable measurements of rain, snow, or fog without mechanical failures or evaporation errors.

WO2024236891A1PCT designated stage expired Publication Date: 2024-11-21MURATA MFG CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
PCT/JP2024/009091
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-18
Filing Date
2024-03-08
Publication Date
2024-11-21

AI Technical Summary

Technical Problem

Conventional weather measuring devices face maintenance issues due to mechanical mechanisms, clogging from debris, and inaccuracies in measuring rain, snow, or fog due to evaporation and sunlight interference, while existing systems cannot detect falling rain, snow, or fog particles effectively.

Method used

A weather measuring device with a housing having an open top, equipped with a light source and an infrared camera for image detection, utilizing machine learning to estimate the amount of falling particles based on image analysis, and incorporating sensors for environmental data to create and update algorithms, eliminating the need for mechanical maintenance and clogging concerns.

Benefits of technology

The device provides a maintenance-free solution for accurately estimating the amount of falling rain, snow, or fog particles, unaffected by clogging or evaporation, using machine learning to correlate image data with precipitation amounts, enhancing precision and reliability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure JP2024009091_21112024_PF_FP_ABST
    Figure JP2024009091_21112024_PF_FP_ABST
Patent Text Reader

Abstract

Provided is a maintenance-free meteorological measurement device capable of accurately estimating precipitation in the form of rainfall, snowfall, or mist without having to worry about the clogging of dust and moisture evaporation. A rain gauge (1) is configured to comprise: a light guide plate unit (3) and an infrared camera (4) in a housing (2); and an AI computer (21) which constitutes an information processing device. The housing (2) has a rectangular parallelepiped shape in which the top surface and the bottom surface are open. The light guide plate unit (3) is provided on one side surface of the housing (2). The light guide plate unit (3) is configured such that an LED (3a), a light guide plate (3b), and a diffusion plate (3c) are housed in a unit case (3d). The infrared camera (4) is provided on another side surface, facing the one side surface via a space (5), of the housing (2). The AI computer (21) constitutes an image detection means for detecting the image of raindrops (6) and a precipitation estimation means for estimating rainfall from the image of the raindrops (6) through machine learning.
Need to check novelty before this filing date? Find Prior Art

Description

Meteorological measuring equipment

[0001] The present invention relates to a meteorological measuring device for measuring the amount of rain, snow or fog particles falling.

[0002] A conventional meteorological measurement device of this type is disclosed, for example, in Patent Document 1. This meteorological measurement device includes a light source, a light-emitting mechanism consisting of a collimating lens that converts light emitted from the light source into parallel light, a condensing lens positioned a certain distance from the collimating lens, and a light-receiving mechanism consisting of an image sensor that receives light that passes through the condensing lens and outputs image data, all mounted on a support. The support includes a roughly U-shaped arm member and a base member that hangs down from the center of the arm member and is installed on the ground or the like. The light-emitting mechanism is mounted in a light-emitting system housing formed at the top end of one vertical bar of the arm member, and the light-receiving mechanism is mounted in a light-receiving system housing formed at the top end of the other vertical bar of the arm member. The base member supports the arm member so that it can rotate. Raindrops fall into the optical path of the parallel light, producing image data in which only the raindrops are depicted as black shadows. Processing this image data allows calculations of rainfall, raindrop size, and raindrop fall speed.

[0003] Another example of a conventional meteorological measurement device of this type is the rain gauge disclosed in Patent Document 2. This rain gauge includes a water collection section that collects rainwater that falls within a water receiving opening, a tubular section that includes a drip section that drips the rainwater collected in the water collection section, a detection section that detects the rainwater dripping from the drip section using infrared rays, and a calculation section that calculates the amount of precipitation based on the detection results.

[0004] Furthermore, a conventional flow rate estimation system for estimating the flow rate of a water current is disclosed in, for example, Patent Document 3. This flow rate estimation system includes an imaging device that captures an image of the water current emitted from a nozzle, and an information processing device that estimates the flow rate of the water current based on the image captured by the imaging device. The information processing device includes an image acquisition unit that acquires an image of the water current, an estimation unit that estimates the flow rate of the water current based on the acquired image through machine learning, and an output unit that outputs the estimation result.

[0005] JP 2008-157765 A JP 2019-164134 A JP 2022-99959 A

[0006] The conventional meteorological measuring device disclosed in Patent Document 1 has a mechanical mechanism with a movable part, in which an arm member equipped with a light-emitting mechanism and a light-receiving mechanism is rotatably supported on a base member. Therefore, when the meteorological measuring device is used, maintenance is required, just like a tipping bucket-type rain gauge that has a tipping bucket as a mechanical mechanism.

[0007] Furthermore, in the rain gauge disclosed in Patent Document 2, if leaves or debris get stuck in the detection section, the infrared light is blocked from reaching the raindrops, making it impossible to accurately determine the amount of rain. Also, because the detection section is configured to detect raindrops that gradually drip from the drip section, when the amount of precipitation is small, the raindrops in the drip section evaporate, making it impossible to accurately determine the amount of rainfall.

[0008] Furthermore, the flow rate estimation system disclosed in Patent Document 3 detects the flow of water emitted from a nozzle, and therefore cannot detect the amount of rain, snow, or fog falling from the sky.

[0009] The present invention has been made to solve these problems, and provides a meteorological measurement device comprising: a housing with an open top; a light source provided on one side of the housing and irradiating light into a space formed within the housing below the top; a photographing device provided on the other side of the housing opposite the one side across the space and photographing the light irradiated from the light source; an image detection means for detecting, from an image photographed by the photographing device, an image of rain, snow or fog particles falling through the space via the top surface blocking the light irradiated from the light source; and an information processing device having a descent amount estimation means for learning, by machine learning, the correlation between the image detected by the image detection means and the measured amount of particle descent, and estimating the amount of particle descent from the image.

[0010] According to this configuration, the mechanical mechanism of the device does not have any moving parts, and therefore, unlike the meteorological measurement device disclosed in the above-mentioned conventional Patent Document 1 or the tipping bucket rain gauge commonly used to measure rainfall, no maintenance is required when operating the device.

[0011] Furthermore, because a large space can be created within the housing below the top surface, there is no risk of leaves or debris clogging the area where raindrops are detected, as in the conventional rain gauge disclosed in Patent Document 2. Furthermore, because the amount of raindrops falling is estimated by machine learning based on the image formed by shining light onto the droplets, there is no risk of raindrops evaporating and the amount of rainfall not being determined correctly, as in the conventional rain gauge disclosed in Patent Document 2.

[0012] This means that the system is maintenance-free, and there is no need to worry about clogging with leaves or debris or evaporation of water, and it can learn and accurately estimate the amount of rain, snow, or fog particles that fall using machine learning.

[0013] According to the present invention, it is possible to provide a meteorological measurement device that is maintenance-free and can accurately estimate the amount of rain, snow, or fog particles that will fall without having to worry about clogging with leaves or debris or evaporation of water.

[0014] 1A is an external perspective view of a rain gauge according to one embodiment of the present invention, and FIG. 1B is a longitudinal cross-sectional view thereof. FIG. 1B is a block diagram of a rain gauge according to one embodiment. FIG. 1C is a general flowchart showing an overview of the precipitation estimation process performed by an AI computer constituting the rain gauge according to one embodiment. FIG. 3 is a flowchart showing details of the raindrop extraction process performed by the image detection means in step 101. FIG. 3 is a flowchart showing the process of detecting the rainfall intensity detected by the raindrop sensor in step 101. FIG. 1D is a table showing the average values ​​per minute of various data obtained by the raindrop extraction process shown in FIG. 4 and the rainfall intensity detection process shown in FIG. 5 in a chronological order. FIG. 1E is a graph showing an example of the correlation between the number of raindrop images and the amount of precipitation obtained by machine learning in the AI ​​computer constituting the rain gauge according to one embodiment.

[0015] Next, an embodiment of the meteorological measuring device of the present invention will be described.

[0016] FIG. 1(a) is a perspective view of the appearance of a rain gauge 1 according to one embodiment in which the meteorological measurement device of the present invention is applied to a rain gauge, and FIG. 1(b) is a longitudinal sectional view thereof.

[0017] The rain gauge 1 is configured by providing a light guide plate unit 3 and an infrared camera 4 in a housing 2 , and providing a single-board computer (not shown) near the housing 2 .

[0018] The housing 2 is a rectangular parallelepiped with an opening at the top and bottom, a width of 320 mm, a depth of 375 mm, and a height of 380 mm, and is made of resin or the like. A light guide plate unit 3 is provided on one side of the housing 2. The light guide plate unit 3 is configured by housing an LED (light emitting diode) 3a, which constitutes a light source, a light guide plate 3b, and a diffusion plate 3c in a unit case 3d.

[0019] Multiple LEDs 3a are provided on the sides of the light guide plate 3b. Each LED 3a emits infrared light in a wavelength range (near-to-far infrared) close to the absorption wavelength of water and where sunlight is significantly attenuated, e.g., near-infrared light with a wavelength of 940 nm. A light guide plate 3b and a diffuser plate 3c are provided on the light-exiting side of each LED 3a. Near-infrared light incident from the side of the light guide plate 3b travels through a light guide path extending throughout the entire interior of the light guide plate 3b, resulting in surface emission from the front surface of the light guide plate 3b, which faces the back surface of the diffuser plate 3c. The diffuser plate 3c diffuses the light emitted from the light guide plate 3b in a planar form, causing it to exit the rectangular opening 2b of the housing 2 as uniform light. Therefore, the near-infrared light emitted from the LEDs 3a passes through the light guide plate 3b and the diffuser plate 3c, irradiating the space 5 formed within the housing 2 below the top surface as near-infrared light. When the raindrops 6 fall into the space 5, they pass through the near-infrared light.

[0020] The infrared camera 4, which constitutes the imaging device, is mounted on one side of the housing 2, opposite the other side across the space 5. The camera detects and visualizes infrared light irradiated into the space 5, and captures near-infrared light irradiated from the LEDs 3a. The housing 2 has a canopy 2a above the infrared camera 4, which protrudes into the space 5, to block raindrops 6 falling into the space 5. In addition, a raindrop sensor 7 is mounted below the space 5 of the housing 2 to measure the amount of rainfall from the falling raindrops 6. This raindrop sensor 7 is only needed when installing the rain gauge 1 and when updating the algorithm, which will be described later.

[0021] FIG. 2 is a block diagram of the rain gauge 1 according to one embodiment.

[0022] The AI ​​computer 21 is a single-board computer such as NVIDIA's Jetson Nano, and is electrically connected to an LED control unit 22 that controls the light emission of the LED 3a, and to the infrared camera 4. The AI ​​computer 21 is also electrically connected to the raindrop sensor 7, wind direction and speed sensor 8, noise sensor 9, vibration sensor 10, illuminance sensor 11, and odor sensor 12. These various sensors are provided in the housing 2, etc., to detect the environment around the rain gauge 1.

[0023] The wind direction and speed sensor 8 detects the wind direction and speed around the housing 2. The noise sensor 9 detects noise occurring around the housing 2 and in the housing 2 itself. The vibration sensor 10 detects vibrations occurring around the housing 2 and in the housing 2 itself. The illuminance sensor 11 detects the illuminance around the housing 2. The odor sensor 12 detects odors around the housing 2. It is not necessary to provide all of these sensors 8 to 12, and sensors may be selected and used as appropriate.

[0024] The AI ​​computer 21 is equipped with a GPU, a CPU, memory, storage, etc., and constitutes an information processing device. The information processing device has an image detection means and a precipitation amount estimation means. The image detection means detects images of raindrops 6 falling through the ceiling and into the space 5, blocking the near-infrared light emitted from the LED 3a, from video images captured by the infrared camera 4. In this embodiment, the image detection means detects images of raindrops 6. The raindrops, snowdrops, or fogdrops absorb near-infrared light, causing the area where the particles are present to appear clearly black on the infrared camera 4. The precipitation amount estimation means uses machine learning to learn the correlation between the image detected by the image detection means and the amount of precipitation of the particles measured as a reference by the raindrop sensor 7, and estimates the amount of precipitation of the particles from the image. In this embodiment, the precipitation amount estimation means estimates the amount of precipitation from the images of the raindrops 6.

[0025] The fall amount estimation means includes an algorithm creation means and an algorithm update means. The algorithm creation means learns, through machine learning, the correlation between the image detected by the image detection means and the measured amount of particle fall, and creates an algorithm used to estimate the amount of particle fall. When any of the sensors 8 to 12 is connected to the AI ​​computer 21, the algorithm creation means takes into account the detection information detected by the sensors 8 to 12 being used and learns, through machine learning, the correlation between the detected image and the measured amount of particle fall, and creates an algorithm. The algorithm update means uses images newly detected by the image detection means to re-learn, through machine learning, the correlation between the particle image and the amount of particle fall, and updates the algorithm. This re-learning of the algorithm is performed by acquiring precipitation data at various locations.

[0026] The descent amount estimation means further includes an image complement processing means. The image complement processing means performs complement processing to mitigate momentary fluctuations in the number, area, or shape of images detected in time series by the image detection means, in post-processing after image detection, based on information on images before and after the fluctuation. The descent amount estimation means uses the images complemented by the image complement processing means to learn the correlation between the particle images and the particle descent amounts through machine learning.

[0027] The image detection means, descent amount estimation means, algorithm creation means, algorithm update means and image complement processing means are realized by software processing in accordance with the program of the AI ​​computer 21.

[0028] 3 is a general flowchart showing an outline of the precipitation amount estimation process performed by the AI ​​computer 21. This precipitation amount estimation process is performed when the rain gauge 1 is installed and whenever the algorithm is updated thereafter.

[0029] First, in step 101, images of raindrops 6 captured by the infrared camera 4 by irradiating the space 5 with near-infrared light from the LED 3a are measured by the image detection means. Measurements are performed at multiple locations and multiple times. The various measured images are stored as a database in the memory of the AI ​​computer 21, along with the amount of precipitation measured by the raindrop sensor 7 and the sensor data detected by the various sensors 8 to 12.

[0030] Next, in step 102, an algorithm for calculating the amount of precipitation from the images of raindrops 6 detected by the image detection means is created by the algorithm creation means through supervised machine learning. At this time, the AI ​​computer 21 automatically extracts rain-related features from the video of the rain, and the algorithm is created based on more diverse information than ever before.

[0031] Next, in step 103, the amount of precipitation is calculated by the rainfall amount estimation means from images of raindrops 6 obtained during a specific period using the database accumulated in step 101 and the algorithm created in step 102. This calculation of the amount of precipitation can be performed in real time.

[0032] Next, in step 104, re-learning is performed by machine learning to which the newly obtained image of the raindrop 6 is added. Next, step 102 is performed again, and an algorithm that reflects the results of the re-learning in step 104 is recreated by the algorithm update means, thereby constantly updating the previously created algorithm. Thereafter, the processes of steps 102, 103, and 104 are repeatedly executed, and the precipitation amount is calculated using the updated algorithm.

[0033] FIG. 4 is a flowchart showing the details of the raindrop extraction process performed by the image detection means in step 101.

[0034] First, in step 201, raw data (.h264) of a video captured by the infrared camera 4 is acquired from a database stored in memory. The video is captured at 100 frames per second with a pixel count of 480 x 320 [px].

[0035] Next, in step 202, frame difference processing is performed on the raw data acquired in step 101 to detect moving images, i.e., images of raindrops 6. Then, in the next step 203, the image data of the detected raindrops 6 is converted into binary data.

[0036] Next, in step 204, the binarized data is subjected to a closing morphological transformation process. In this closing process, an expansion process is performed on the image of the raindrop 6, which is the object, and then an erosion process is performed to fill in the small black dots contained in the object.

[0037] Next, in step 205, object boundary detection processing is performed on the binarized data that has been subjected to the closing processing, and images of raindrops 6 are detected. Next, in step 206, processing is performed to calculate the number of images of raindrops 6 in each frame, i.e., the number of droplets.

[0038] Next, in step 207, the raindrop count data for each frame is saved in a pickle file, and the raindrop counts are stored in chronological order. At the same time, the sensor data detected at each timing by the wind direction and speed sensor 8, noise sensor 9, vibration sensor 10, illuminance sensor 11, and odor sensor 12 are also stored in the database in chronological order along with the raindrop counts, in accordance with the timing of the detection of the raindrop 6 images.

[0039] Next, in step 208, a resampling process is performed to match the sampling rate of raindrops 6 in the rainfall intensity sensor data measured from the raindrop sensor 7 as described below with the sampling rate of the images of raindrops 6 extracted in the raindrop extraction process. The video data, captured at 100 frames per second by the infrared camera 4, produces one frame of image every 0.01 seconds. In the resampling process, the average number of raindrop images 6 appearing per frame is calculated, for example, per 1 minute, 5 minutes, 30 minutes, or 60 minutes, in accordance with the calculation rate of the raindrops 6 in the rainfall intensity sensor data. Furthermore, in accordance with the resampling rate of the number of raindrops, the average value per frame (per 0.01 seconds) of the information detected per minute, 5 minutes, 30 minutes, or 60 minutes is calculated for the sensor data from the various sensors 8 to 12, for example, in accordance with the resampling rate of the number of raindrops.

[0040] 5 is a flowchart showing the process of detecting the rainfall intensity detected by the raindrop sensor 7. This rainfall intensity detection process is also performed in step 101 of FIG.

[0041] First, in step 301, the AI ​​computer 21 acquires raw data (.csv) obtained at each timing from the raindrop sensor 7. This raw data is expressed in 0 to 123 steps of voltage from 0 to 5 V.

[0042] Next, in step 302, the rainfall intensity is calculated based on the sensor specifications of the raindrop sensor 7 and the voltage obtained in step 301 on a scale of 0 to 123.

[0043] Next, in step 303, a resampling process is performed on the rainfall intensity (amount of precipitation) obtained in step 302. In this resampling process, an average value per frame of the rainfall intensity detected per minute, 5 minutes, 30 minutes, or 60 minutes is calculated in accordance with the sampling rate of the images of raindrops 6 extracted in the raindrop extraction process.

[0044] 6 is a table showing, in time series, the average values ​​per minute of various data obtained by the raindrop extraction process shown in Fig. 4 and the rainfall intensity detection process shown in Fig. 5, with the granularity of the image data of the raindrops 6 and the rainfall intensity data adjusted by the resampling processes 208 and 303. These data are stored as a database in the memory of the AI ​​computer 21.

[0045] In the figure, the sensor data of illuminance [lx], vibration [dB], and wind speed [m / s] obtained by the illuminance sensor 11, vibration sensor 10, and wind direction and speed sensor 8, as well as the number of detected raindrops [pieces], which is the number of images of raindrops 6 detected in the raindrop extraction process, and the amount of precipitation [mm / h] detected in the rainfall intensity detection process are shown in a table for each time.

[0046] Based on such data accumulated in the processing of step 101 in Figure 3, the precipitation amount estimation means embodied by the AI ​​computer 21 learns, through machine learning in processing of step 102, the correlation between the image of the raindrop 6 detected by the image detection means and the rainfall intensity (precipitation amount) of the raindrop 6 measured by the raindrop sensor 7, and estimates the precipitation amount of the raindrop 6 from the image in processing of step 103.

[0047] FIG. 7 is a graph showing an example of the correlation between the number of images of raindrops 6 and the amount of precipitation obtained by this machine learning.

[0048] The horizontal axis of the graph is the number of detected raindrops 6 (detected particle number [pieces / frame]), and the vertical axis is the rain intensity (rain intensity [mm / hour]) obtained from the output of the raindrop sensor 7. In addition, the plots marked with stars (★) and crosses (+) represent the detected data of the images used in machine learning, with the plots marked with stars representing the observed train data and the plots marked with crosses representing the observed test data.

[0049] Furthermore, curve 31 represents the mean of the learning model function, which expresses the correlation between the number of detected particles and rainfall intensity, obtained from the learning data, and region 32, which is represented by a width centered on characteristic line 31, represents the reliability of the data. In areas where there is a large distribution of learning data and the number of detected particles on the horizontal axis is small, the width of region 32 is represented as small, indicating high reliability of the data. In areas where there is a small distribution of learning data and the number of detected particles on the horizontal axis is large, the width of region 32 is represented as large, indicating low reliability of the data.

[0050] The rainfall intensity prediction using machine learning is performed by learning the correlation between the images of raindrops 6 detected by the image detection means and the amount of precipitation at the time of detection using machine learning, and then performing a regression analysis. In this embodiment, the regression analysis is performed using Gaussian process regression, which obtains the probability distribution of a learning model function representing the correlation between the number of detected raindrops and rainfall intensity in the form of a Gaussian process. In addition, in the Gaussian process regression, both the RBF kernel and the linear kernel are used as kernel functions. Note that the rainfall intensity prediction using machine learning is not limited to Gaussian process regression, and may also be performed using image classification architectures such as VGG (Visual Geometry Group) and ResNet (Residual Networks).

[0051] As shown in the graph of FIG. 7, a correlation between the number of detected raindrops and the rainfall intensity is obtained, and from the number of detected raindrops 6 obtained by the image detection means in a specific period, the amount of precipitation is calculated by the precipitation amount estimation means based on the curve 31 in step 103 of FIG. 3.

[0052] According to the rain gauge 1 of this embodiment, the mechanical mechanism of the device does not have any moving parts, and therefore, unlike the conventional meteorological measurement device disclosed in Patent Document 1 or the tipping bucket rain gauge commonly used to measure rainfall, no maintenance is required when operating the device.

[0053] Furthermore, because the space 5 formed within the housing 2 below the top surface can be made large, there is no risk of leaves or debris clogging the area where raindrops are detected, as in the conventional rain gauge disclosed in Patent Document 2. Furthermore, because the amount of raindrops 6 falling is estimated by machine learning based on the images formed by shining light onto the droplets, there is no risk of raindrops evaporating and the amount of rainfall not being determined correctly, as in the conventional rain gauge disclosed in Patent Document 2.

[0054] This makes it possible to learn the amount of falling raindrops 6 through machine learning and accurately estimate it without any maintenance or concerns about clogging with leaves or debris or evaporation of water.

[0055] The rain gauge disclosed in Patent Document 2 can only count the number of raindrops falling from the dripping part, but the rain gauge 1 according to this embodiment uses image data to detect images of raindrops 6, so it is possible to detect various information such as the number, size, and falling speed of raindrops 6 and correlate this information with the amount of precipitation to predict the amount of precipitation. Also, the meteorological measurement device disclosed in Patent Document 1 can extract characteristics of the size, falling speed, and shape of raindrops, but cannot correlate this information with the amount of precipitation.

[0056] Furthermore, in the rain gauge 1 according to this embodiment, the amount of falling raindrops 6 is estimated using an algorithm created by the algorithm creating means using rainfall data acquired under various conditions in step 102 of Fig. 3. Then, in step 104, the algorithm is re-learned by the algorithm updating means using new images detected by the image detecting means. This allows the amount of falling raindrops 6 to be estimated with even greater accuracy.

[0057] Furthermore, with the rain gauge 1 according to this embodiment, non-visual information such as wind direction and speed, noise, vibration, and odor detected around the housing 2 or on the housing 2 itself, as well as illuminance information, are added to the learning data of the machine learning performed by the algorithm creation means, and the correlation between the image detected by the image detection means and the amount of raindrop 6 falling can be estimated. This makes it possible to predict rainfall taking into account changes in the surrounding conditions that could not be taken into account before, and estimates the correlation between the image of the raindrop 6 and the amount of raindrop 6 falling in various environments, allowing for accurate estimation of this correlation in each environment, thereby enabling more accurate estimation of the amount of raindrop 6 falling in each environment. This makes it possible to predict rainfall that takes into account differences in location, time, and the influence of the surroundings.

[0058] As a result, rain-related characteristics can be automatically extracted from video footage of rain using machine learning, making it possible to calculate precipitation using a wider variety of information than ever before. For example, when the wind blows, the falling direction of raindrops 6 and the number of observed raindrops 6 also change, but it was previously unknown how these characteristics relate to the amount of precipitation. However, according to this embodiment, by using machine learning, it is possible to mechanically associate the information on the falling direction and number of raindrops 6 captured in the video with the amount of precipitation.

[0059] Furthermore, in the rain gauge 1 according to this embodiment, a post-processing step after image detection, called an image complementing step, is carried out by the image complementing means based on information on the images before and after the change, to mitigate momentary fluctuations in the number, area, or shape of the images detected in time series by the image detecting means. The correlation between the image and the amount of precipitation is learned by the rainfall amount estimating means using the images complemented by the image complementing means.

[0060] For example, even if a momentary strong wind reduces the number of raindrops 6 falling into the space 5 of the rain gauge 1, the sensor data from the wind direction and speed sensor 8 and the image data of the raindrops 6 are referenced in chronological order, and the image count data of the raindrops 6 detected when the wind speed increases can be supplemented based on information on images before and after the change, thereby correcting the estimated amount of precipitation when the wind speed increases. Therefore, the correlation between the images and the amount of precipitation can be accurately estimated by the precipitation amount estimation means.

[0061] Furthermore, with the rain gauge 1 according to this embodiment, by using infrared rays in a wavelength band close to the absorption wavelength of water to detect the images of the raindrops 6, most of the infrared rays emitted from the LEDs 3a are absorbed by the raindrops 6, and the images of the raindrops 6 captured by the infrared camera 4 become clearer. Furthermore, by using infrared rays in a wavelength band in which sunlight is largely attenuated to detect the images of the raindrops 6, the effects of sunlight shining on the housing 2 can be reduced and the images of the raindrops 6 can be captured.

[0062] Therefore, with this configuration, it is possible to capture clear images of raindrops 6 with little influence from sunlight. On the other hand, the meteorological measurement device disclosed in Patent Document 1 is strongly influenced by sunlight and other external light, and is therefore unable to accurately determine the amount of rain, etc.

[0063] Furthermore, in the rain gauge 1 according to this embodiment, the light emitted from the LED 3a is converted into planar light by the light guide plate 3b and then converted into uniform planar light by the diffuser plate 3c. Therefore, the light emitted from the LED 3a is evenly irradiated into the space 5 formed within the housing 2 below the top surface. Therefore, the light is evenly irradiated onto the raindrops 6 falling into the space 5 between the light guide plate unit 3 and the infrared camera 4, and the infrared camera 4 captures images of the raindrops 6 without missing any. Therefore, the image detection means can accurately detect the images of the raindrops 6, and thus the correlation between the images of the raindrops 6 and the amount of raindrops 6 that fall can be accurately estimated.

[0064] The meteorological measurement device disclosed in Patent Document 1 requires an expensive telecentric lens as the parallelizing lens, which increases the price of the meteorological measurement device, but the rain gauge 1 according to this embodiment can be realized using a general light guide plate 3b and diffusion plate 3c, thereby reducing the product price.

[0065] Furthermore, with the rain gauge 1 according to this embodiment, the canopy 2a above the infrared camera 4 prevents raindrops 6 from falling on the infrared camera 4. This makes it possible to eliminate the influence of raindrops 6 adhering to the infrared camera 4 on the image captured by the infrared camera 4.

[0066] In the above embodiment, the present invention has been described as being applied to a rain gauge 1 that detects images of raindrops 6 to predict and estimate precipitation. However, the present invention is not limited to raindrops 6, and can also be applied to meteorological measurement devices that detect snow or fog particles in the same way as raindrops 6 to predict and estimate the amount of snowfall or fog. In this case, the same effects as those of the rain gauge 1 according to the above embodiment can be achieved.

[0067] To summarize the above, the present invention can be expressed as follows.

[0068] <1> A meteorological measurement device comprising: a housing with an open top; a light source provided on one side of the housing and irradiating light into a space formed within the housing below the top; an imaging device provided on the other side of the housing facing the one side across the space and imaging the light irradiated from the light source; an image detection means for detecting, from an image captured by the imaging device, an image of rain, snow, or fog particles falling through the space via the top surface blocking the light irradiated from the light source; and an information processing device having a descent amount estimation means for learning, by machine learning, a correlation between the image detected by the image detection means and a measured descent amount of the particles, and estimating the descent amount of the particles from the image.

[0069] <2> The meteorological measurement device described in <1>, characterized in that the precipitation amount estimation means has: an algorithm creation means that learns the correlation through machine learning and creates an algorithm used to estimate the precipitation amount of the particles; and an algorithm update means that uses the image newly detected by the image detection means to re-learn the correlation through machine learning and update the algorithm.

[0070] <3> The meteorological measurement device described in <2> is equipped with at least one of a wind direction and speed sensor that detects wind direction and speed around the housing, a noise sensor that detects noise occurring around the housing and in the housing itself, a vibration sensor that detects vibrations occurring around the housing and in the housing itself, an odor sensor that detects odors around the housing, and an illuminance sensor that detects illuminance around the housing, and is characterized in that the algorithm creation means creates the algorithm by learning the correlation through machine learning, taking into account detection information detected by at least one of the sensors.

[0071] <4> The meteorological measurement device described in any one of <1> to <3>, characterized in that the descent amount estimation means is provided with an image completion processing means that performs completion processing to mitigate momentary fluctuations in the number, area, or shape of the images detected in time series by the image detection means in post-processing after the detection of the images, based on information on the images before and after the fluctuations, and the correlation is learned by machine learning using the images that have been completion-processed by the image completion processing means.

[0072] <5> The meteorological measurement device described in any one of <1> to <4>, characterized in that the light source emits infrared rays in a wavelength band close to the absorption wavelength of water and where attenuation of sunlight is large, and the photographing device is an infrared camera that detects and visualizes the infrared rays.

[0073] <6> A meteorological measuring device described in any one of <1> to <5>, characterized in that it is provided with a light guide plate that causes the light emitted from the light source to emit light in a surface-emitting manner, and a diffusion plate that diffuses the light emitted from the light guide plate on the light emission side of the light source.

[0074] <7> A meteorological measurement device described in any one of <1> to <6>, characterized in that the housing has a canopy above the imaging device that protrudes into the space to prevent the particles from falling into the space.

[0075] DESCRIPTION OF SYMBOLS 1...Rain gauge (meteorological measurement device) 2...Housing 2a...Eaves 2b...Rectangular opening 3...Light guide plate unit 3a...LED (light source) 3b...Light guide plate 3c...Diffusion plate 3d...Unit cover 4...Infrared camera 5...Space 6...Raindrop 7...Raindrop sensor 8...Wind direction and speed sensor 9...Noise sensor 10...Vibration sensor 11...Illuminance sensor 12...Smell sensor 21...AI computer (information processing device) 22...LED control unit CROSS-REFERENCE TO RELATED APPLICATIONS

[0076] This application claims priority based on Japanese Patent Application No. 2023-082558, filed with the Japan Patent Office on May 18, 2023, the entire disclosure of which is incorporated herein by reference in its entirety.

Claims

1. A meteorological measurement device comprising: a housing with an open top; a light source provided on one side of the housing and irradiating light into a space formed within the housing below the top; an imaging device provided on the other side of the housing facing the one side across the space and imaging the light irradiated from the light source; an image detection means for detecting an image of rain, snow or fog particles falling through the space via the top surface blocking the light irradiated from the light source from an image captured by the imaging device; and an information processing device having a descent amount estimation means for learning by machine learning a correlation between the image detected by the image detection means and a measured descent amount of the particles, and estimating the descent amount of the particles from the image.

2. The meteorological measurement device described in claim 1, characterized in that the precipitation amount estimation means comprises: an algorithm creation means which learns the correlation through machine learning and creates an algorithm used to estimate the amount of precipitation of the particles; and an algorithm update means which uses the image newly detected by the image detection means to re-learn the correlation through machine learning and update the algorithm.

3. The meteorological measuring device of claim 2, further comprising at least one of a wind direction and speed sensor that detects wind direction and speed around the housing, a noise sensor that detects noise occurring around the housing and in the housing itself, a vibration sensor that detects vibrations occurring around the housing and in the housing itself, an odor sensor that detects odors around the housing, and an illuminance sensor that detects illuminance around the housing, wherein the algorithm creation means creates the algorithm by learning the correlation through machine learning, taking into account detection information detected by at least one of the sensors.

4. The meteorological measurement device described in any one of claims 1 to 3, characterized in that the descent amount estimation means is equipped with an image completion processing means that performs completion processing to mitigate momentary fluctuations in the number, area, or shape of the images detected in time series by the image detection means, in post-processing after the detection of the images, based on information on the images before and after the fluctuations, and the correlation is learned by machine learning using the images that have been completed by the image completion processing means.

5. A meteorological measuring device as claimed in any one of claims 1 to 4, characterized in that the light source emits infrared rays in a wavelength band close to the absorption wavelength of water and where attenuation of sunlight is large, and the imaging device is an infrared camera that detects and visualizes infrared rays.

6. A meteorological measuring device as described in any one of claims 1 to 5, characterized in that a light guide plate that emits light emitted from the light source in a surface-emitting manner and a diffusion plate that diffuses the light emitted from the light guide plate are provided on the light emission side of the light source.

7. A meteorological measuring device as described in any one of claims 1 to 6, characterized in that the housing has a canopy above the imaging device that protrudes into the space to prevent the particles from falling into the space.

Citation Information

Patent Citations

  • Apparatus and method for sensing snowfall

    JP1996201534A

  • Detecting device for snow-coverage on road surface

    JP1998213674A

  • Weather measurement apparatus

    JP2010032557A

  • Fallout detection device

    JP2010236932A

  • Precipitation information processing system, precipitation information processing method, precipitation information processing program, and unmanned aircraft

    JP2021060376A