WO2012005461A2

WO2012005461A2 - Method for automatically calculating information on clouds

Info

Publication number: WO2012005461A2
Application number: PCT/KR2011/004628
Authority: WO
Inventors: 차주완; 장기호; 이철규; 권중장; 최영진
Original assignee: 대한민국(기상청장)
Priority date: 2010-07-09
Filing date: 2011-06-24
Publication date: 2012-01-12
Also published as: KR20120005654A; WO2012005461A3; KR101183105B1

Abstract

The present invention relates to a method for automatically calculating information on clouds, wherein the method comprises: a first step of inputting a video image of the atmosphere including a cloud area; a second step of extracting the cloud area from the input video image; a third step of calculating the amount and heights of the clouds; a fourth step of analyzing the forms of the clouds through a cloud pattern using the calculated amount and heights of the clouds; and a fifth step of calculating the forms of the clouds for the upper, middle and lower clouds. According to the present invention, the forms of the clouds for the upper, middle and lower clouds can be automatically calculated using the amount and heights of the clouds which are calculated by analyzing the video image including the cloud image, thereby automatically obtaining objective observation data of the clouds for each level. The method for automatically calculating information on clouds according to the present invention can maximize observation efficiency through the automation of cloud observation and secure objective data with high resolution on the height and amount of the clouds which are uncertain, thereby significantly enhancing the technology of weather forecasting and climate change prediction.

Description

Automatic cloud information calculation method

The present invention relates to a method for automatically calculating data on cloudiness, cloud height, cloud shape for each cloud in the upper, middle, and lower layers of the sky.

In the world of meteorological and climatic observations, the condition of the sky (clouds) has generally been performed to the eye. However, the observation of retrospective meteorological observations has been interrupted, resulting in a problem that the spatial resolution of the observation cloud observation data is degraded. In addition, weather workers and local government office workers have difficulty in performing observations such as cloudiness, cloud shape, cloud height, and meteorological phenomena on the three-hour interval.

The meteorological observation does not present a specific meteorological calculation algorithm at all in terms of methodological method using cloud and cloud information.In particular, the meteorological zones of the world are currently used for visual observation according to WMO (World Meteorological Organization) standards. The situation is recorded by observing rhythm. Therefore, there is a need to automatically calculate and record the cloud shape of the upper, middle and lower layer clouds using the cloud height calculated by the cloud image method.

The present invention has been made to solve the above problems, an object of the present invention is to automatically calculate the cloud shape of the upper cloud, middle cloud, lower cloud by using the cloud amount, cloud height value calculated by analyzing the image image including the cloud image In addition, the present invention provides a method of calculating cloud information to obtain objective observation data of automated layered clouds.

As a means for solving the above problems, the present invention comprises the steps of inputting a video image of the atmosphere including the cloud area; Extracting a cloud region from the input image image; Calculating cloud and cloud height using the cloud area; Analyzing cloud patterns by analyzing cloud patterns using the calculated clouds and clouds; And five steps of calculating upper, middle, and lower stratus rhythms.

In this case, the first step may be configured to acquire and input an image image of the atmosphere including the cloud area by using two left and right cameras.

In addition, the second step may be a step of extracting only the cloud area by dividing the cloud area and the sky area by using the change or contrast of the color image in the video image.

In addition, in the third step, after classifying the cloud area and the non-cloud area for each image pixel in the extracted cloud area, only the cloud area is crystallized to derive the percentage (%) of only the cloud area in the entire image. Cloud level calculation step for each pixel; and cloud level calculation step for each pixel for deriving the cloud height by determining the height of the cloud area of the corresponding position of the left and right image images obtained by the two left and right cameras.

In this case, the cloud-by-pixel calculation step, when there is one cloud layer, derives the average height of the pixel areas as the cloud height observed at that time, and classifies the classification in the observed image having a plurality of cloud layers. Divided into middle and lower cloud height can be implemented to derive the average cloud by floor.

In addition, step 4 may be formed by analyzing cloud patterns by comparing cloud cloud lookup table information forming a cloud shape database optimized for regional characteristics with the observed image.

More specifically, step 4 may include: a1) converting the pixel image of the observation area into a vector format; A2) classifying the similar cloud pattern through the neural network analysis method using the information generated in step a1; A3) generating cloud-optimized lookup tables optimized for regions by classifying the cloud patterns classified in step a2) into 10 types of cloud types based on WMO cloud cloud standards; Comparing the cloud cloud form lookup table optimized for each region and cloud image images observed in real time to calculate the best cloud shape; may be configured to include.

The system configuration for implementing the above-described calculation method can be implemented in the following configuration.

Specifically, the automatic cloud information system according to the present invention comprises an image input unit for inputting the image of the atmosphere including the cloud area; A cloud region extraction unit for extracting a cloud region from the input image; A cloud cloud computing unit for calculating cloud clouds and cloud height using the cloud area; It may be configured to include; cloud pattern analysis unit for analyzing the cloud shape by analyzing the cloud pattern calculated by the cloud cloud height calculation unit.

In addition, the image input unit may be configured as an image input unit consisting of two cameras, left and right, to provide an image image of the observation point.

In addition, the cloud cloud computing unit, the pixel-specific cloud computing unit for crystallizing only the image of the cloud region in the input image image by the percentage (%) of only the cloud region in the entire image image; and the cloud region in the video image And a cloud-by-pixel computing unit for calculating an average cloud height by unrolling the bay.

In addition, the cloud pattern analysis unit, by comparing the cloud cloud pattern lookup table optimized for the observation area and the actual cloud pattern can be compared to calculate the cloud shape for each of the upper, middle, and lower layers.

According to the present invention, by automatically calculating the cloud shape of the upper cloud, the middle cloud, the lower cloud by using the cloud amount, cloud height value calculated by analyzing the image image including the cloud image, an automated observation of the objective cloud data of the layered cloud There is.

Due to the automatic cloud information calculation method according to the present invention, it is possible to maximize the observation efficiency by cloud observation automation, weather forecasting and climate change prediction technology by securing high-resolution objective data on the height and quantity of clouds with high uncertainty There is also an effect that can be significantly improved.

1 is a conceptual diagram of a cloud observation information calculation system of the upper and lower middle clouds in the sky state according to the present invention.

2 is a block diagram showing the configuration of the automatic cloud observation system 100 according to the present invention.

3 is a flowchart illustrating a method for calculating automatic cloud information according to the present invention.

4 is a conceptual diagram illustrating a process of forming a cloud cloud form look-up table.

Hereinafter, with reference to the accompanying drawings will be described in detail the configuration and operation according to the present invention. In the description with reference to the accompanying drawings, the same components are given the same reference numerals regardless of the reference numerals, and duplicate description thereof will be omitted. Terms such as first and second may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another.

The present invention is an automatic floor-by-bed (upper, middle, lower layer) by integrated automatic observation and calculation of the cloud (cloud) conditions performed by the eye in the field of meteorological and climate observation worldwide, cloud, cloud, cloud form The objective is to provide a method and system for producing objective observation data of clouds.

1 is a conceptual diagram of a cloud observation information calculation system of the upper and lower middle clouds in the sky state according to the present invention. In the present invention, it is possible to remotely manage the cloudiness, cloud height, cloud shape occurring locally by using the correlation between cloud and cloud shape. In particular, the automatic cloud observation system 100 according to the present invention for calculating the optimal information of the cloud shape by comparing the image input unit 10 for automatic observation of the cloud image, and the normalized cloud pattern information and the observed cloud information carried out It characterized by having a.

Referring to the drawings, the automatic cloud observation system according to the present invention includes an image input unit 10 for inputting the image of the atmosphere including the cloud region and the cloud region extraction unit 110 for extracting the cloud region from the input image, Cloud pattern analysis unit 130 and the cloud pattern analysis unit 130 that calculates the cloud and cloud height by using the cloud area, cloud pattern analysis unit 130 to analyze the cloud pattern calculated by the cloud cloud cloud computing unit 120 and the final It is configured to include a cloud shape calculation unit 140 for calculating the cloud shape.

The image input unit 10 may be configured as an image input unit consisting of two cameras, left and right, to provide an image image of an observation point.

In addition, the cloudiness cloud computing unit 120 crystallizes only the cloud region image in the image image input from the image input unit 10 to form a percentage (%) of only the cloud region in the entire image image. It may be configured to include a calculation unit 121 and the pixel-specific cloud computing unit 122 for calculating the average cloud height by clouding only the cloud area in the image image. For cloud and cloud observation, two cameras are used, and color cloud images are used during the day and contrast is measured at night.

The cloud pattern analysis unit 130 performs a function of calculating cloud patterns for upper, middle, and lower layers by comparing and analyzing the cloud cloud lookup table 150 optimized for the observation area and the cloud patterns observed in reality. Based on the observed cloudiness and cloudage information, cloud image pattern analysis (PCA (Principle Component Analysis), PNN (Probailistic Neural Network, etc.) enables the integrated calculation of cloudiness, cloudiness, and cloud shape. In order to classify cloud types, cloud image patterns of long-term upper, middle and lower stratum cloud images should be defined in consideration of cloud volume and cloud height. It can be generated and automatically calculate the cloud shape considering the cloud quantity, cloud height, etc. of the cloud image observed in real time.

A method of calculating integrated information of cloud volume, cloud height, and cloud shape using the functions of each component will be described with reference to FIG. 3.

Referring to the flowchart shown, the automatic cloud information calculation method according to the present invention comprises the first step of inputting the image image of the atmosphere including the cloud region, the second step of extracting the cloud region from the input image image, and the cloud Three steps of calculating the cloud and cloud by using the region, the cloud pattern is analyzed using the cloud and cloud calculated by using the cloud pattern comprises four steps to analyze the cloud shape and five steps to calculate the upper, middle, lower cloud rhyme do.

The first step is a step of acquiring an atmospheric image using the left and right cameras as the image input unit according to the present invention and inputting it into the system, and the second step is extracting only the wrong region from the input image. In this case, it is preferable to extract only the cloud area by eliminating the effects of the basic preprocessing for estimating the cloud, such as the sun and the moon.

The third step is to calculate the cloud amount of only the cloud area, which is performed by the pixel-specific cloud computing unit in the above-described system. Specifically, in order to distinguish the cloud area from the sky area in the acquired image image, the change of the color image is used during the day, and the contrast is used at night. In order to estimate the cloudiness, basic pre-processing (remove effect by sun, moon, etc.), display the cloud part and non-cloud part by image pixel, and then crystallize only the cloud area to change the amount of cloud area only in% in full screen. The cloud is calculated to calculate the cloud volume (pixel level calculation step).

Subsequently, in the cloud-by-pixel calculation step for each cloud, only the cloud area of the acquired image image is clouded and the average height is calculated. This pixel-specific cloud height is determined by using two cameras to determine the height of the cloud region at the corresponding position, and define the height as the cloud height of the image region. At this time, the cloud height for each pixel is divided into histograms to distinguish whether a cloud is a single layer or a multilayer. If there is a cloud layer, the average height of these pixel areas is the cloud height observed at that time. If there are two or more cloud layers, the observations are classified in the observation image and divided into lower and middle clouds or upper clouds, and the average cloud is calculated separately. This is used as the basic data for classifying rhymes. In order to verify the results of these observations, the existing observation system is compared with other cilometers (Ceilometer). In other words, after the cloud amount and cloud height are calculated, whether a single cloud or a multi-layer cloud may be classified.

Four steps of analyzing the cloud pattern by the image and statistical processing according to the upper, middle, lower cloud classification made in the above three steps may be performed. In this step 4, it is desirable to build a zonal classification system and DB optimized for the region so that the rhythm classification can be automated.

For example, it is desirable to form sufficient observation data to database the cloud form by classifying cloud form according to WMO cloud classification criteria. That is, it is necessary to secure observation data of cloud type by upper, middle, and lower cloud classification.

The cloud pattern analysis method for cloud classification can be performed as follows.

That is, the cloud pattern analysis method for cloud classification has statistical methods for various image processing, and it is preferable to select and apply the optimal method for the region to be observed. To analyze these cloud patterns, it can be divided into the process of forming a look-up table optimized for the area and the process of recognizing cloud type by comparing the cloud image with the real-time observed cloud image. .

This process can be described as follows with reference to FIG.

Step 1: The process of selecting a technical algorithm to create a cloud cloud lookup table is performed. In this case, a suitable method such as PCA (Principle Component Analysis) or SVM (Super Vector Machine) is used to create the most suitable cloud lookup table for each region. The basis of all these methods is that the conversion of each observation pixel image into a vector form must be done first.

Step 2: Classify similar cloud patterns through neural network analysis of cloud images generated by the method selected in step 1 to create cloud cloud lookup table. In analyzing cloud patterns, cloud patterns are analyzed in consideration of cloud volume and cloud height to improve cloud classification classification accuracy.

Step 3: The classified cloud patterns are classified into 10 types of cloud types, which are the WMO cloud cloud standard, to generate a cloud cloud survey table optimized for the region.

-Step 4: Based on the cloud cloud lookup table optimized for each region, the cloud patterns are analyzed in real time by the same method applied in step 1, and the calculated upper, middle, lower cloud and cloud height are considered. Compare with cloud cloud form chart. By performing this comparison process repeatedly, the best rhyme is calculated.

In summary, the automatic cloud information calculation method according to the present invention integrates cloud, cloud and cloud type by analyzing cloud image pattern (PCA (Principle Component Analysis), PNN (Probailistic Neural Network, etc.) based on observed cloud and cloud information. In particular, to classify cloud patterns that occur locally, the cloud image patterns of long-term upper, middle, and lower stratum clouds are normalized by considering the cloud volume and cloud height, and the cloud characteristics of each altitude are defined. The cloud form can be automatically calculated by considering the cloudiness, cloud height, etc. of the cloud image observed in real time. Although it was a method of calculating, in the present invention, cloudiness, cloudiness, and cloud form are organically combined and automatically calculated.

In the detailed description of the invention as described above, specific embodiments have been described. However, many modifications are possible without departing from the scope of the invention. The technical spirit of the present invention should not be limited to the described embodiments of the present invention, but should be determined not only by the claims, but also by those equivalent to the claims.

Claims

Inputting a video image of an atmosphere including a cloud area;

Extracting a cloud region from the input image image;

Calculating cloud and cloud height using the cloud area;

Analyzing cloud patterns by analyzing cloud patterns using the calculated clouds and clouds; And

5 steps to calculate the upper, middle, lower rhyme;

Automatic cloud information calculation method comprising a.
The method according to claim 1,

The first step,

An automatic cloud information calculating method comprising obtaining and inputting an image of an air including a cloud area using two left and right cameras.
The method according to claim 2,

The second step,

Automatic cloud information calculation method comprising the step of extracting only the cloud region by separating the cloud region and the sky region using the change or contrast of the color image in the image image.
The method according to claim 3,

The third step,

Cloud-by-pixel cloud computing step of deriving a percentage (%) of only the cloud area in the entire image by crystallizing only the cloud area after distinguishing the cloud area and the non-cloud area for each image pixel with respect to the extracted cloud area; Wow

A cloud-by-pixel calculation step for deriving a cloud height by determining a height of a cloud area at a corresponding position of the left and right image images obtained by the two left and right cameras;

Automatic cloud information calculation method composed of.
The method according to claim 4,

The pixel-specific cloud height calculation step,

If there is one cloud layer, the average height of the pixel regions is derived from the cloud height observed at that time, and the average cloud level is divided into upper, middle, and lower cloud heights. Automatic cloud information calculation method, characterized in that deriving.
The method according to claim 4,

The fourth step,

Comparing cloud cloud lookup table information forming a cloud shape database optimized for the regional characteristics and the observed image image is a step of analyzing the cloud pattern automatic cloud information calculation method.
The method according to claim 6,

The fourth step,

A1 step of converting the pixel image of the observation area into a vector format;

A2) classifying the similar cloud pattern through the neural network analysis method using the information generated in step a1;

A3) generating cloud-optimized lookup tables optimized for regions by classifying the cloud patterns classified in step a2) into 10 types of cloud types based on WMO cloud cloud standards;

A4) calculating the best cloud form by comparing cloud cloud form lookup tables optimized for each region with cloud image images observed in real time;

Automatic cloud information calculation method comprising a.
An image input unit which inputs an image of the atmosphere including a cloud area;

A cloud region extraction unit for extracting a cloud region from the input image;

A cloud cloud computing unit for calculating cloud clouds and cloud height using the cloud area;

A cloud pattern analysis unit for analyzing cloud patterns by analyzing cloud patterns calculated by the cloud cloud computing unit;

Automatic cloud information system comprising a.
The method according to claim 8,

The image input unit comprises an image input unit consisting of two cameras, the left and right, automatic cloud information calculation system, characterized in that to provide a video image of the observation point.
The method according to claim 9,

The cloud cloud computing unit,

Cloud-by-pixel calculation unit that crystallizes only the image of the cloud region in the input image image to make a percentage (%) of only the cloud region in the entire image image; and

A cloud-by-pixel calculation unit for calculating an average cloud height by clouding only the cloud area in the video image;

Automatic cloud information calculation system comprising a.
The method according to claim 10,

The cloud pattern analysis unit,

Automatic cloud information calculation system, characterized in that the cloud cloud form lookup table optimized for the observation area and the cloud pattern observed in the real world is compared and analyzed to calculate the cloud shape for each of the upper, middle, and lower layers.