KR102389960B1 - Sea fog or fog detection system and method using satellite photos - Google Patents

Abstract

A system and method for detecting sea fog or fog using satellite imagery are provided. Sea fog or fog detection system according to an embodiment of the present invention, the communication unit for acquiring satellite information, weather station information and CCTV image information; and a processor that detects sea fog or fog based on the acquired data. Accordingly, sea fog or fog is detected based on satellite information, weather station information and CCTV image information, visibility distance is calculated based on the detection result, and road fog, rainfall, snowfall, snow, It is possible to identify and provide the current state of freezing, and it is possible to provide a real-time nationwide fog map service.

Description

Sea fog or fog detection system and method using satellite photos

The present invention relates to a system and method for detecting sea fog or fog, and more particularly, to a system and method for detecting sea fog or fog using satellite images.

Fog on land and at sea plays a very important role in human life. In the case of land, in general, it is closely related to real life, such as causing traffic accidents, and when combined with pollutants such as smog, it can have a very large effect on public health. In the sea, fog is directly related to the safety of ships and also has a great effect on the fishing industry. In addition, various human casualties can occur due to fog.

Satellites are the only means of real-time, wide-area fog detection. In places where there are many human activities such as cities, ground observation such as observation is possible, but there is a problem that such direct observation is limited and very difficult in places such as remote areas or the sea.

Most of the existing satellite-based fog detection techniques are based on true color, which is a combination of spectral channels corresponding to red-green-blue colors, which are similar to those observed with the human eye. Black-and-white images that depend on images or observed from satellites have been mainly used. However, fog detection using artificial satellites is a difficult technology, and it is difficult to distinguish between low-level clouds (lower clouds) and fog. In addition, if there is another cloud above the fog, it is difficult for the satellite to detect the fog in the lower layer, and the false positive rate is also very high.

On the other hand, the conventional visibility distance calculation technique is the maximum distance that can clearly identify the target. The degree of atmospheric turbidity was determined, and in the past it was limited to fog and yellow dust, but its importance is growing as the causes of occurrence such as fine dust, large fires, and abnormal climates are increasing in modern times.

In addition, fog detection technology is mainly human observation data, and the observation value lacks consistency. Recently, a detection method using satellite data has been developed, but it has limitations in distinguishing low-level clouds from fog. Existing forward scattering type visibility systems have a problem in that accurate measurement is difficult because light is not reflected in an environment where yellow dust and fine dust are generated.

Korean Patent Registration No. 10-1986025 (Title of Invention: Machine Learning-based Satellite Sea Fog Detection Device and Sea Fog Detection Method)

The present invention has been devised to solve the above problems, and an object of the present invention is to detect sea fog or fog based on satellite information, weather station information, and CCTV image information, and calculate a visibility distance based on the detection result It is to provide a sea fog or fog detection system.

According to an embodiment of the present invention for achieving the above object, a system for detecting sea fog or fog includes: a communication unit for acquiring satellite information, weather station information, and CCTV image information; and a processor that detects sea fog or fog based on the acquired data.

And the satellite information includes date information, time information, lower cloud image information, night fog image information, infrared image information, visible ray image information, image patch information and pixel information, and the weather station information includes weather information around the weather station. and station identification information.

In addition, the image patch information is generated by dividing the satellite image into a plurality of regions, matching a patch unique number for each divided region, and the processor, low clouds based on weather station information and each image patch information, It is determined whether sea fog or fog is detected, and if low cloud, sea fog or fog is detected in a specific area, the CCTV video information in the area where low cloud, sea fog or fog is detected is searched. It can be judged that haze has occurred.

And the processor searches for CCTV image information in the area where sea fog or fog is detected, and when CCTV image information is found, based on the retrieved CCTV image information, it can be determined whether fog or low clouds have occurred in the area. there is.

In addition, the processor performs deep learning learning using a convolutional neural network, deconvolution, and skip connection for each of the image patch information and CCTV image information obtained from the satellite. can do.

And the processor selects image patch information corresponding to within a preset radius based on the weather station, and based on the weather information around the weather station, generates a feature map for applying attention for detecting sea fog or fog And, using the generated feature map, deep learning learning can be performed.

Also, the processor may derive the longitude and latitude of a region corresponding to the image patch information based on information on pixels constituting each image patch information.

In addition, the processor may calculate a visibility distance for each region corresponding to the image patch information based on information on pixels constituting each image patch information.

Also, the processor may image-process pixels constituting each image patch information using a convolutional neural network, and may calculate a viewing distance based on a result of the image processing.

Meanwhile, according to another embodiment of the present invention, a method for detecting sea fog or fog includes: acquiring, by a sea fog or fog detection system, satellite information, weather station information, and CCTV image information; and detecting, by the sea fog or fog detection system, the sea fog or fog based on the acquired data.

As described above, according to embodiments of the present invention, sea fog or fog is detected based on satellite information, weather station information, and CCTV image information, a visibility distance is calculated based on the detection result, and a road CCTV image is used. It is possible to identify and provide the current state of fog, rainfall, snowfall, snow accumulation, and icing on the road, and it is possible to provide a real-time nationwide fog map service.

1 is a view provided for the description of a sea fog or fog detection system using a satellite image according to an embodiment of the present invention;
2 is a view provided for explaining the configuration of a sea fog or fog detection system using a satellite image according to an embodiment of the present invention;
3 is a diagram provided for explanation of image patch information;
4 is a diagram illustrating satellite image information;
5 is a view provided for explaining a deep learning learning method using a sea fog or fog detection system using a satellite image according to an embodiment of the present invention, and
6 to 9 are views provided to explain a method for detecting sea fog or fog using a system for detecting sea fog or fog using a satellite image according to an embodiment of the present invention.

Hereinafter, the present invention will be described in more detail with reference to the drawings.

1 is a view provided for explanation of a sea fog or fog detection system 100 (hereinafter, collectively referred to as 'sea fog or fog detection system 100') using a satellite image according to an embodiment of the present invention; 2 is a diagram provided to explain the configuration of the sea fog or fog detection system 100 according to an embodiment of the present invention, FIG. 3 is a diagram provided to describe image patch information, and FIG. It is an illustrated drawing.

The sea fog or fog detection system 100 according to this embodiment acquires satellite information including satellite images from the Meteorological Agency 10, and weather information around the weather station 20 from the weather stations 20 installed here and there. Acquire the included weather station 20 information, and acquire CCTV image information from the CCTV 30 or CCTV video management center installed in an arbitrary area, based on satellite information, weather station 20 information, and CCTV image information It is possible to detect sea fog or fog, calculate a visibility distance based on the detection result, and provide the fog, rainfall, snowfall, snow, and icing status of the road based on the CCTV image of the road.

To this end, the present sea fog or fog detection system 100 may include a communication unit 110 , a processor 120 , an output unit 130 , and a storage unit 140 .

The communication unit 110 may receive or transmit data while the processor 120 operates.

Specifically, the communication unit 110 may acquire satellite information, weather station 20 information, and CCTV image information.

For example, obtain satellite information including satellite images from the Meteorological Agency 10, and obtain weather station 20 information including weather information around the weather station 20 from weather stations 20 installed in various places. , it is possible to acquire CCTV image information from a CCTV installed in an arbitrary area or a CCTV video management center.

Here, the satellite information may include date information, time information, infrared image information, water vapor image information, RGB day/night composite image information, image patch information, and pixel information, and the weather station 20 information includes the weather station 20 ) may include the surrounding weather information and the observatory 20 identification information.

In this case, FIG. 4A is a diagram exemplifying infrared image information, FIG. 4B is a diagram exemplifying water vapor image information, and FIG. 4C is a diagram exemplifying RGB day/night composite image information.

The processor 120 detects sea fog or fog based on the data obtained through the communication unit 110, calculates a visibility distance based on the detection result, and based on the road CCTV image, fog, rain, snow, You can check the snow and ice conditions.

To this end, the processor 120, for each of the image patch information and CCTV image information obtained from the satellite, a convolutional neural network (convolutional neural network), deconvolution (deconvolution) and a dip using a skip connection (skip connection) learning can be performed.

That is, the processor 120 selects image patch information corresponding to within a preset radius based on the weather station 20, and based on the weather information around the weather station 20, attention ( A feature map for applying attention) can be generated, and deep learning can be performed using the generated feature map.

In addition, the processor 120 may generate image patch information for detecting sea fog or fog by dividing the satellite image into a plurality of regions, and matching a unique patch number for each of the divided regions.

Specifically, the processor 120 may generate image patch information for each image information when infrared image information, water vapor image information, and RGB day/night composite image information are obtained among the satellite images.

In addition, the processor 120 divides the infrared image information, the water vapor image information, and the RGB day and night composite image information, respectively, and uses the image patch information generated three for each zone, and uses the low cloud, sea fog or It is possible to determine whether fog is detected and to calculate the visibility distance for each area.

For example, the processor 120 determines whether low cloud, sea fog or fog is detected based on the weather station 20 information and each image patch information, and when low cloud, sea fog or fog is detected in a specific area, If the CCTV image information is not retrieved by searching for CCTV image information in the area where low clouds, sea fog or fog are detected, it can be determined that sea fog has occurred in the corresponding area.

In addition, the processor 120 searches for CCTV image information in the area in which sea fog or fog is detected, and when CCTV image information is found, based on the retrieved CCTV image information, whether fog or low clouds have occurred in the area can determine whether

Through this, when low cloud, fog, or sea fog is detected through the satellite image, it is possible to determine whether the detected target is low cloud, fog, or sea fog, thereby minimizing the detection error of low cloud, fog, or sea fog. .

And the processor 120 derives the longitude and latitude of a region corresponding to the image patch information based on the information of the pixels constituting each image patch information, and based on the information of the pixels constituting the respective image patch information It is possible to calculate the visibility distance for each area corresponding to the image patch information.

The output unit 130 is a display device provided to output data to the screen when the processor 120 operates, and the storage unit 140 stores programs and data necessary for the processor 120 to operate. is a storage medium.

5 is a diagram provided to explain a deep learning learning method using the sea fog or fog detection system 100 according to an embodiment of the present invention.

In the deep learning learning method using the sea fog or fog detection system 100 according to an embodiment of the present invention, when satellite information, weather station 20 information and CCTV image information are acquired (S510), the satellite image is displayed in a plurality of zones. , and it is possible to generate image patch information corresponding to each of the divided regions (S520).

Thereafter, the deep learning learning method using the sea fog or fog detection system 100 selects image patch information corresponding to within a preset radius based on the weather station 20 ( S530 ), and the weather around the weather station 20 . Based on the information, a feature map for applying an attention for detecting sea fog or fog may be generated (S540), and deep learning may be performed using the generated feature map (S550).

Here, the sea fog or fog detection system 100 generates a feature map for detecting sea fog or fog based on the selected image patch information when image patch information corresponding to within a preset radius is selected based on the weather station 20 . and applying attention to an area necessary for detecting sea fog or fog according to the feature map to filter out unnecessary parts.

6 to 9 are views provided for explanation of a method for detecting sea fog or fog using the sea fog or fog detection system 100 according to an embodiment of the present invention.

The sea fog or fog detection method using the sea fog or fog detection system 100 according to an embodiment of the present invention performs deep learning learning according to the method described above with reference to FIG. 5 , and then again, satellite information, weather station (20) When information and CCTV image information are obtained (S610), the satellite image may be divided into a plurality of regions, and image patch information corresponding to each of the divided regions may be generated (S620).

Next, the sea fog or fog detection method using the sea fog or fog detection system 100 may determine whether low clouds, sea fog, or fog are detected based on the weather station 20 information and each image patch information (S630) ).

In addition, the sea fog or fog detection method using the sea fog or fog detection system 100 may calculate the visibility distance for each area corresponding to the image patch information based on information on pixels constituting each image patch information.

Here, FIG. 7 is a diagram illustrating a process of detecting sea fog or fog by analyzing CCTV image information and satellite photos (image information), and FIG. 8 is a diagram illustrating a process of analyzing CCTV image information. And FIG. 9 is a diagram illustrating an algorithm for digitizing the visibility distance for each area by performing a convolutional neural network, deconvolution, and skip connection for each image information am.

For example, as illustrated in FIG. 9 , the sea fog or fog detection system 100 designates a region of interest according to metadata necessary for sea fog or fog detection in information of pixels constituting each image patch information, By filtering regions not included in the region, the visibility distance for each region can be calculated using a convolutional neural network, deconvolution, and skip connection that generates a plurality of convolutional layers. can

Through this, sea fog or fog is detected based on satellite information, weather station 20 information, and CCTV image information, visibility distance is calculated based on the detection result, and fog, rainfall, and snowfall on the road are based on the road CCTV image. , snow cover, and freezing conditions can be identified and provided, and real-time nationwide fog map service can be provided.

On the other hand, it goes without saying that the technical idea of the present invention can be applied to a computer-readable recording medium containing a computer program for performing the functions of the apparatus and method according to the present embodiment. In addition, the technical ideas according to various embodiments of the present invention may be implemented in the form of computer-readable codes recorded on a computer-readable recording medium. The computer-readable recording medium may be any data storage device readable by the computer and capable of storing data. For example, the computer-readable recording medium may be a ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical disk, hard disk drive, or the like. In addition, the computer-readable code or program stored in the computer-readable recording medium may be transmitted through a network connected between computers.

In addition, although preferred embodiments of the present invention have been illustrated and described above, the present invention is not limited to the specific embodiments described above, and the technical field to which the present invention belongs without departing from the gist of the present invention as claimed in the claims In addition, various modifications are possible by those of ordinary skill in the art, and these modifications should not be individually understood from the technical spirit or perspective of the present invention.

10: Meteorological Agency
20: Weather Station
30 : CCTV
100: Sea fog or fog detection system
110: communication department
120 : processor
130: output unit
140: storage

Claims (10)

A communication unit for acquiring satellite information, weather station information, and CCTV image information; and
A processor for detecting sea fog or fog based on the acquired data;
satellite information,
including date information, time information, infrared image information, water vapor image information, RGB day/night composite image information, image patch information, and pixel information;
Weather station information,
contains weather information around the weather station and station identification information;
Image patch information,
The satellite image is divided into a plurality of regions, and a patch unique number is matched for each divided region to generate,
The processor is
Determine whether low clouds, sea fog or fog are detected based on weather station information and each image patch information;
When low cloud, sea fog or fog is detected in a specific area, the CCTV image information in the area where low cloud, sea fog or fog is detected is searched.
The processor is
For each of the image patch information and CCTV image information obtained from the satellite, using a convolutional neural network, deconvolution and skip connection,
The processor is
Selecting image patch information within a preset radius based on a weather station, and generating a feature map for applying attention for sea fog or fog detection based on weather information around the weather station,
Sea fog or fog detection system, characterized in that performing deep learning learning by using the generated feature map.
delete delete The method according to claim 1,
The processor is
Sea fog, characterized in that by searching for CCTV image information in an area where sea fog or fog is detected, and when CCTV image information is retrieved, based on the retrieved CCTV image information, it is determined whether there is fog or low cloud in the area or fog detection systems.
delete delete The method according to claim 1,
The processor is
A system for detecting sea fog or fog, characterized in that the longitude and latitude of a region corresponding to the image patch information are derived based on information on pixels constituting each image patch information.
8. The method of claim 7,
The processor is
A system for detecting sea fog or fog, characterized in that the visibility distance for each area corresponding to the image patch information is calculated based on information on pixels constituting each image patch information.
9. The method of claim 8,
The processor is
A system for detecting sea fog or fog, comprising image processing pixels constituting each image patch information using a convolutional neural network, and calculating a visibility distance based on a result of the image processing.
A method for detecting sea fog or fog, comprising: acquiring satellite information, weather station information, and CCTV image information; and
Including, by the sea fog or fog detection system, detecting sea fog or fog based on the acquired data;
satellite information,
including date information, time information, infrared image information, water vapor image information, RGB day/night composite image information, image patch information, and pixel information;
Weather station information,
contains weather information around the weather station and station identification information;
Image patch information,
The satellite image is divided into a plurality of regions, and a patch unique number is matched for each divided region to generate,
The step of detecting sea fog or fog,
Determine whether low clouds, sea fog or fog are detected based on weather station information and each image patch information;
When low cloud, sea fog or fog is detected in a specific area, the CCTV image information in the area where low cloud, sea fog or fog is detected is searched.
Sea fog or fog detection system,
For each of the image patch information and CCTV image information obtained from the satellite, using a convolutional neural network, deconvolution and skip connection,
Sea fog or fog detection system,
Selecting image patch information within a preset radius based on a weather station, and generating a feature map for applying attention for sea fog or fog detection based on weather information around the weather station,
Sea fog or fog detection method, characterized in that performing deep learning learning using the generated feature map.

Images