KR102569735B1 - Multi-satellite performance evaluation method and system - Google Patents

Abstract

The present invention preprocesses the first satellite image and the second satellite image so that the shooting ranges overlap, specifies at least a part of the region in each image, and detects an object from the specified region of the first satellite image by using a previously prepared object detection algorithm. A first entity distribution image is extracted to reveal the detected location, and a second entity distribution image is extracted to reveal the location where the object is detected from a specified area of the second satellite image, and the first entity distribution image and the second entity distribution image are extracted. Provided is a multi-satellite performance evaluation method that compares object distribution images and calculates a similarity of at least one of object distribution patterns and density to be used for performance evaluation between satellites.

Description

Multi-satellite performance evaluation method and system {MULTI-SATELLITE PERFORMANCE EVALUATION METHOD AND SYSTEM}

The present invention relates to a multi-satellite performance evaluation method and system.

Marine algae grow by attaching themselves to the bottom of shallow waters, some of which rise to the surface and move with currents. Coastal waters are rich in nutrients supplied from the land or farms, so green algae (spiny blue) and brown algae (Hawsaeng's cap) reproduce in large quantities even while floating on the sea surface, and can be easily observed on satellite images.

In this regard, as a method of detecting algae through satellite images, a method of detecting algae on the surface of seawater using a difference in reflectivity in the near infrared and visible light regions by applying an index for measuring terrestrial vegetation to the satellite image is used. In addition, recently, methods for detecting algae such as green algae and brown algae floating on the surface of the sea water have been used in consideration of changing environments such as the shooting angle of satellite images, atmospheric conditions, and turbidity of sea water.

Meanwhile, the satellite may perform missions such as communication with ground stations, earth observation, weather, navigation and positioning, and space science research. At this time, the ground station is a device or software of the earth that communicates with a satellite located in space, and based on this, the satellite enters orbit by a launch vehicle, acquires exploration data such as satellite images, and transmits them to the ground station.

Accordingly, by using the information obtained from the satellite, it is possible to study the cycle of water and energy, changes in the ocean, chemical reactions in the atmosphere, the surface of the earth, and polar ice by observing the earth's weather and surface. Furthermore, by using satellite images of the ocean, it is possible to continuously monitor the marine environment over a wide area, such as the concentration of phytoplankton in seawater, the concentration of suspended sediment, the concentration of dissolved organic matter, and the distribution of floating algae.

However, since such a satellite directly visits the satellite orbit located in space, on-site maintenance and collection are almost impossible, based on the information observed from the satellite, whether the satellite is performing well in its observation mission or A method for quantitatively evaluating whether performance is being maintained is required.

The present invention relates to a method and system for evaluating the performance of a satellite by comparing a result of detecting an object in a satellite image with a result according to a satellite image captured by another satellite.

In addition, the present invention relates to a multi-satellite performance evaluation method and system for evaluating similarity by analyzing and comparing at least one of distribution patterns and densities of the same entity in different satellite images.

In order to solve the problems described above, a multi-satellite performance evaluation method according to the present invention includes the steps of inputting a first satellite image and a second satellite image different from the first satellite image; performing pre-processing on at least one of the first satellite image and the second satellite image, and specifying at least a portion of an area in each of the first satellite image and the second satellite image so that a photographing range overlaps; A first object distribution image is extracted to reveal a location where an object is detected from a specified area of the first satellite image by using a previously prepared object detection algorithm, and a specified object of the second satellite image is extracted using the object detection algorithm. extracting a second object distribution image to reveal a location where the object is detected from the region; And comparing the first entity distribution image and the second entity distribution image to be used for performance evaluation between a first satellite capturing the first satellite image and a second satellite capturing the second satellite image, The method may include calculating a similarity of at least one of distribution patterns and densities of objects detected in each of the first satellite image and the second satellite image.

In addition, the multi-satellite performance evaluation system according to the present invention includes a communication unit to which a first satellite image and a second satellite image different from the first satellite image are input; and performing preprocessing on at least one of the first satellite image and the second satellite image to specify at least a portion of an area in each of the first satellite image and the second satellite image so that a photographing range overlaps, and A first object distribution image is extracted to reveal a location where an object is detected from a specified area of the first satellite image using an object detection algorithm, and a object distribution image is extracted from a specified area of the second satellite image using the object detection algorithm. A second entity distribution image is extracted to reveal a location where an object is detected, and the first entity distribution image and the second entity distribution image are compared to reveal a location where the object is detected, and a first satellite capturing the first satellite image and the second entity distribution image and a control unit that calculates a similarity of at least one of a distribution pattern and a density of objects detected in each of the first satellite image and the second satellite image so as to be used for performance evaluation between the second satellites that have captured the satellite image. can

In addition, the program stored in the computer-readable recording medium according to the present invention is executed by one or more processes in the electronic device and is stored in the computer-readable recording medium, and the program is a first satellite inputting an image and a second satellite image different from the first satellite image; performing pre-processing on at least one of the first satellite image and the second satellite image, and specifying at least a portion of an area in each of the first satellite image and the second satellite image so that a photographing range overlaps; A first object distribution image is extracted to reveal a location where an object is detected from a specified area of the first satellite image by using a previously prepared object detection algorithm, and a specified object of the second satellite image is extracted using the object detection algorithm. extracting a second object distribution image to reveal a location where the object is detected from the region; And comparing the first entity distribution image and the second entity distribution image to be used for performance evaluation between a first satellite capturing the first satellite image and a second satellite capturing the second satellite image, Calculating a similarity of at least one of distribution patterns and densities of objects detected in each of the first satellite image and the second satellite image;

According to various embodiments of the present invention, the multi-satellite performance evaluation system detects each object from a plurality of satellite images and calculates the similarity of the distribution pattern and density of the object based on the detection result, thereby confirming the difference in performance between the satellites. can be used to do

In addition, according to various embodiments of the present invention, a multi-satellite performance evaluation method and system determines the distribution pattern and density of objects for a satellite image taken by another satellite based on a satellite image taken by one satellite. It can be used to determine the performance level of the satellite by calculating the similarity to the satellite.

1 shows a multi-satellite performance evaluation system according to the present invention.
2 is a flowchart illustrating a multi-satellite performance evaluation method according to the present invention.
3 to 5 illustrate a process of specifying at least a portion of an area in each of a plurality of satellite images.
6 is a flowchart illustrating a method of extracting an entity distribution image from a satellite image.
7 illustrates an embodiment of an object distribution image.
8 to 10 show an embodiment of calculating a similarity for a distribution pattern.
11 illustrates an embodiment of calculating a degree of similarity for a degree of density.

Hereinafter, the embodiments disclosed in this specification will be described in detail with reference to the accompanying drawings, but the same reference numerals will be assigned to the same or similar components regardless of reference numerals, and overlapping descriptions thereof will be omitted. The suffixes "module" and "unit" for components used in the following description are given or used together in consideration of ease of writing the specification, and do not have meanings or roles that are distinct from each other by themselves. In addition, in describing the embodiments disclosed in this specification, if it is determined that a detailed description of a related known technology may obscure the gist of the embodiment disclosed in this specification, the detailed description thereof will be omitted. In addition, the accompanying drawings are only for easy understanding of the embodiments disclosed in this specification, the technical idea disclosed in this specification is not limited by the accompanying drawings, and all changes included in the spirit and technical scope of the present invention , it should be understood to include equivalents or substitutes.

Terms including ordinal numbers, such as first and second, may be used to describe various components, but the components are not limited by the terms. These terms are only used for the purpose of distinguishing one component from another.

It is understood that when an element is referred to as being "connected" or "connected" to another element, it may be directly connected or connected to the other element, but other elements may exist in the middle. It should be. On the other hand, when an element is referred to as “directly connected” or “directly connected” to another element, it should be understood that no other element exists in the middle.

Singular expressions include plural expressions unless the context clearly dictates otherwise.

In this application, terms such as "comprise" or "have" are intended to designate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, but one or more other features It should be understood that the presence or addition of numbers, steps, operations, components, parts, or combinations thereof is not precluded.

1 shows a multi-satellite performance evaluation system according to the present invention.

Referring to FIG. 1 , the multi-satellite performance evaluation system 100 according to the present invention may collect a plurality of satellite images captured by a plurality of satellites 1a and 1b.

Accordingly, the multi-satellite performance evaluation system 100 detects the same object in each satellite image, generates object distribution images, and compares each object distribution image to determine the distribution pattern and A similarity to at least one of the densities may be calculated.

In this way, the multi-satellite performance evaluation system 100 is used to evaluate the difference in performance between the different satellites 1a and 1b through the similarity calculated for at least one of the distribution patterns and densities of objects detected in the satellite images. Also, it may be used to evaluate the performance of another satellite 1b based on a satellite image captured through one of the satellites 1a.

Furthermore, the multi-satellite performance evaluation system 100 may be used to verify the object detection capability and accuracy of the individual satellites 1 based on the corresponding similarities between the different satellites 1a and 1b, and furthermore, each other It may also be used to set a direction for merging satellite information of other satellites 1a and 1b.

Meanwhile, the plurality of satellites 1a and 1b may be satellites 1 having different performance. Specifically, the plurality of satellites 1a and 1b are configured to capture satellite images at different spatial resolutions. may have been prepared.

For example, one of the plurality of satellites 1a and 1b, as GOCI-2, takes a satellite image with a spatial resolution of 250 m over a range of 2,500 Km X 2,500 Km, and the other, as Landsat OLI, It can capture satellite images over a range of 185 Km X 180 Km with a spatial resolution of 30 m.

Meanwhile, the plurality of satellites evaluated by the multi-satellite performance evaluation system 100 according to the present invention is not limited to the above embodiment, and satellites having differences in various performance and characteristics in addition to spatial resolution may be used. .

In addition, objects detected in the satellite image may refer to floating algae, floating matter, marine debris, spilled oil, etc. whose distribution changes according to marine and meteorological phenomena. For example, an entity detected in satellite imagery may be Sargassum Horneri.

The object distribution image may include information related to an object detected in a satellite image, and the satellite image in which the object is detected may be converted into a preset coordinate system (eg, EPSG 4326) based on latitude and longitude coordinates. Accordingly, the object distribution image may include information related to the location of objects detected from the satellite image and the density of the objects.

The distribution pattern is a pattern in which objects detected in the satellite image are distributed, and may indicate locations where objects exist in the object distribution image. Accordingly, calculating the similarity of distribution patterns using different satellite images may be calculating the similarity of positions of objects detected from different satellite images.

Density is the degree to which objects detected in the satellite image are dense, and may represent a histogram related to the distribution of objects by density in the object distribution image. Therefore, calculating the similarity of the density using different satellite images may be calculating the similarity of the histogram according to the density of objects detected in the different satellite images.

To this end, the multi-satellite performance evaluation system 100 may include a storage unit 110, a communication unit 130, and a control unit 150.

The storage unit 110 may store data and commands necessary for controlling the overall operation of the multi-satellite performance evaluation system 100 . In addition, the storage unit 110 may store satellite images, and may store data generated in the process of calculating a similarity for at least one of a distribution pattern and a density of satellite images.

The communication unit 130 may be connected to the satellite 1 or an external server storing satellite images through a wireless or wired network. Through this, the controller 150 may receive satellite images from the satellite 1 or an external server in which satellite images are stored.

The controller 150 may control overall operations of the multi-satellite performance evaluation system 100 .

Based on the configuration of the multi-satellite performance evaluation system 100 reviewed above, the multi-satellite performance evaluation method will be described in more detail below.

2 is a flowchart illustrating a multi-satellite performance evaluation method according to the present invention. 3 to 5 illustrate a process of specifying at least a portion of an area in each of a plurality of satellite images. 6 is a flowchart illustrating a method of extracting an entity distribution image from a satellite image. 7 illustrates an embodiment of an object distribution image. 8 to 10 show an embodiment of calculating a similarity for a distribution pattern. 11 illustrates an embodiment of calculating a degree of similarity for a degree of density.

Referring to FIG. 2 , when a plurality of satellite images are input, the multi-satellite performance evaluation system 100 performs pre-processing on each of the plurality of satellite images to overlap at least a portion of each of the plurality of satellite images so that the shooting range overlaps. areas of can be specified.

At this time, the multi-satellite performance evaluation system 100 may set one of a plurality of satellite images as a reference image and perform preprocessing on other images so as to overlap the reference image.

For example, the multi-satellite performance evaluation system 100 receives a first satellite image and a second satellite image different from the first satellite image (S100), for at least one of the first satellite image and the second satellite image. Pre-processing may be performed to specify at least a part of the region in each of the first satellite image and the second satellite image so that the photographing ranges overlap (S200).

More specifically, the multi-satellite performance evaluation system 100 may transform a plurality of satellite images into the same coordinate system using information included in each of the plurality of satellite images. Specifically, the multi-satellite performance evaluation system 100 may convert the first satellite image and the second satellite image into a preset coordinate system so that a comparison of the shooting range between the first satellite image and the second satellite image is possible.

For example, the multi-satellite performance evaluation system 100 calculates latitude and longitude coordinates of each of a plurality of pixels included in the satellite image by using latitude and longitude coordinates (Degree) at a plurality of vertices appearing from the satellite image, Latitude and longitude coordinates calculated for each of the plurality of pixels may be converted into a preset coordinate system.

As another example, the multi-satellite performance evaluation system 100, when the satellite image includes the altitude and position of the satellite at the time the satellite image was captured, and the latitude and longitude coordinates at the upper left and lower right corners of the satellite image, Using the altitude and position of the satellite, the latitude and longitude coordinates at the upper left and lower right are converted into the GOES (Geostationary Operational Environmental Satellites) coordinate system, and based on this, for each of a plurality of pixels included in the satellite image, the GOES system Latitude and longitude coordinates can be calculated. Accordingly, the multi-satellite performance evaluation system 100 may convert latitude and longitude coordinates of the GOES system for each of a plurality of pixels into an EPSG 4326 coordinate system as a preset coordinate system.

As another example, the multi-satellite performance evaluation system 100 includes latitude and longitude coordinates corresponding to the four vertexes of the satellite image, EPSG 52361 coordinates, which are a coordinate system different from a preset coordinate system, and spatial resolution, in the satellite image. In this case, latitude and longitude coordinates for each of a plurality of pixels included in the satellite image may be calculated using EPSG 52361 coordinates and spatial resolution, and the calculated latitude and longitude coordinates may be converted into an EPSG 4326 coordinate system as a preset coordinate system. there is.

Furthermore, the multi-satellite performance evaluation system 100 overlaps a satellite image having a large capture range among a plurality of satellite images with a satellite image having a small capture range among a plurality of satellite images so that the capture ranges of the plurality of satellite images overlap. range can be extracted.

In other words, the multi-satellite performance evaluation system 100 may extract a partial range from one of the first satellite image and the second satellite image having a larger capturing range so as to overlap the capturing range of the other satellite image. To this end, a capturing range included in one of the first satellite image and the second satellite image having a smaller capturing range may be the same as at least a part of a capturing range included in the other one.

That is, overlapping the capturing ranges may indicate an area including coordinates of a satellite image having a smaller capturing range in a satellite image having a larger capturing range. That is, the overlapping of the photographing ranges may mean areas in which the photographing ranges correspond or are the same in different satellite images.

For example, among the first satellite image and the second satellite image, the multi-satellite performance evaluation system 100 determines, in the first satellite image, the second satellite image included in the second satellite image when the first satellite image has a larger capturing range. A range including a plurality of pixels corresponding to coordinates of the plurality of pixels may be extracted.

As another example with reference to FIG. 3 , the multi-satellite performance evaluation system 100 determines that the first satellite image 10 has a larger capturing range among the first satellite image 10 and the second satellite image 20 . For example, based on the coordinates of the plurality of pixels included in the first satellite image 10, a plurality of pixels in the first satellite image 10 are mapped so that the coordinates of four corner pixels of the second satellite image 20 are included. It can be extracted with a grid-like range (11).

Furthermore, the multi-satellite performance evaluation system 100 sets a satellite image having a large size among a plurality of satellite images to correspond to a size of a satellite image having a small size among other satellite images so that the sizes of the plurality of satellite images are the same. can be scaled down

In other words, the multi-satellite performance evaluation system 100 may reduce a larger size of the first satellite image and the second satellite image to correspond to the size of the other satellite image.

Alternatively, the multi-satellite performance evaluation system 100 selects a satellite image having a high spatial resolution among a plurality of satellite images and a satellite image having a low spatial resolution among other satellite images so that the spatial resolution of the plurality of satellite images becomes the same. Interpolation may be performed to correspond to the resolution.

For example, in the multi-satellite performance evaluation system 100, when the spatial resolution of the first satellite image is 250 m and the spatial resolution of the second satellite image is 30 m, the spatial resolution of the second satellite image is 250 m. Interpolation can be performed as much as possible.

As another example, the multi-satellite performance evaluation system 100 determines that the size of the second satellite image is 700 X 800 when the size of the first satellite image is 700 X 800 and the size of the second satellite image is 7781 X 7641. Interpolation can be performed so that

As another example with reference to FIG. 4 , when the multi-satellite performance evaluation system 100 reduces the satellite image 30 having a size of 6 X 6 at a ratio of 2 X 3, which is an integer ratio, the satellite image ( 30), an average value of pixels included in the interpolation range 31 of 2 × 3 is sequentially calculated, and an interpolated satellite image 40 having a size of 3 × 2 may be generated.

As another example with reference to FIG. 5 , when the multi-satellite performance evaluation system 100 reduces the satellite image 30 having a size of 6 X 6 at a ratio of 2 X 1.5, which is a real ratio rather than an integer ratio, An interpolated satellite image 50 having a size of 3 X 4 may be generated by sequentially calculating an average value of pixels included in the interpolation range 33 of 2 X 1.5 from one vertex of the satellite image 30. there is.

At this time, the multi-satellite performance evaluation system 100 multiplies the pixel value by the number of decimal places for the pixel corresponding to the number of decimal places among the plurality of pixels included in the satellite image 30 to obtain a range 33 according to the reduction ratio. The average value for can be calculated.

That is, the multi-satellite performance evaluation system 100 calculates (0 + Through the operation of (1 X 0.5) + 6 + (7 X 0.5)), the average value of 3.3 can be calculated.

As another example, the multi-satellite performance evaluation system 100 may adjust the size of a satellite image by using cv2.INTER_AREA (area interpolation) among the resize functions (cv2.resize) included in Python opencvlibrary.

Referring back to FIG. 2 , the multi-satellite performance evaluation system 100 extracts an object distribution image to reveal a location where an object is detected from a region specified for each of a plurality of satellite images using a previously prepared object detection algorithm. can

Specifically, the multi-satellite performance evaluation system 100 extracts a first object distribution image to reveal a location where an object is detected from a specified area of the first satellite image using an object detection algorithm, and uses an object detection algorithm to A second object distribution image may be extracted to reveal a location where an object is detected from a specified area of the second satellite image (S300).

Here, the object detection algorithm may be an algorithm prepared to detect an object from a satellite image and indicate the location and density of the detected object.

Referring to FIG. 6, for example, the multi-satellite performance evaluation system 100 corrects the molecular scattering reflectance in a specified area of the satellite image (S310), and detects clouds and land in the specified area of the satellite image. A preset mask is applied to remove related pixels (S320), and based on preset conditions to extract pixels likely to be determined as objects, among a plurality of pixels included in a specific area of the satellite image, an expected pixel Can be extracted (S330).

Here, correcting the molecular scattering reflectance is to convert the radiance of each pixel observed in the Top Of Atmosphere (TOA) into reflectance by considering the distance between the Earth and the sun, and reflectance by atmospheric molecular scattering (Rayleigh Reflectance ) may be subtracted.

Also, the preset mask may include a cloud pixel mask and a land pixel mask. That is, the multi-satellite performance evaluation system 100 may remove values of pixels corresponding to clouds and the land from a specified area of the satellite image using a preset mask.

At this time, the reference value for the wavelength of the cloud pixel mask is set smaller than the wavelength of the B1 channel (center wavelength: 412 nm, bandwidth: 20 nm) or B3 channel (center wavelength: 490 nm, bandwidth: 20 nm) appearing from the satellite image. and the reference values for the wavelengths of the terrestrial pixel mask are channel B8 (center wavelength: 865 nm, bandwidth: 40 nm), channel B5 (center wavelength: 660 nm, bandwidth: 20 nm), and channel B4 (center wavelength: 555 nm). , bandwidth: 20 nm).

In addition, the pre-set condition for extracting the expected pixel is that the difference between the remote sensing reflectance of the red (Band Red) wavelength in the remote sensing reflectance of the near-infrared (BNir, Band Near-Infrared) is applied to both left and right (or top and bottom) pixels. It may be set to extract as an expected pixel when there is a pixel that is smaller than a preset threshold and the change is not accompanied by a difference in turbidity (Bred).

Furthermore, the multi-satellite performance evaluation system 100 calculates background reflectivity for seawater pixels in a specified area of the satellite image (S340), and calculates the background reflectivity of each of a plurality of pixels included in the specified area of the satellite image. Anomalous spatial reflectivity is calculated by removing the reflectivity (S350), and a pixel having an abnormal spatial reflectivity greater than a preset spatial reflectivity in the noise and visible light bands may be extracted as an object (S360).

Here, the background reflectivity may include an average and standard deviation of reflectivities in a predetermined range (eg, 5 X 5 pixel range) for seawater pixels.

Thereafter, the multi-satellite performance evaluation system 100 may calculate an occupied area of the pixel extracted as an object by considering background reflectivity (S370).

Referring to FIG. 7 , for example, the multi-satellite performance evaluation system 100 generates an object distribution image 60 including information related to locations of objects detected from satellite images and the degree of concentration of objects through an object detection algorithm. can create At this time, the object distribution image 60 may indicate the location of objects detected from a specified area of the satellite image and the degree to which the objects are concentrated through pixel values.

On the other hand, the multi-satellite performance evaluation system 100 may use different algorithms according to the type of object to be detected from satellite images, and furthermore, the multi-satellite performance evaluation system 100 uses a conventional object detection algorithm An object distribution image in which an object is detected may be generated from a satellite image.

In addition, the multi-satellite performance evaluation system 100 may generate object distribution images in which objects are detected using different algorithms prepared to detect the same object in different ways from a plurality of satellite images.

Referring back to FIG. 2, the multi-satellite performance evaluation system 100 compares a plurality of object distribution images and detects them from a plurality of satellite images to be used for performance evaluation between a plurality of satellites that have taken the plurality of satellite images. A similarity of at least one of distribution patterns and densities of the identified entities may be calculated.

Specifically, the multi-satellite performance evaluation system 100 compares the first entity distribution image and the second entity distribution image, and determines between a first satellite capturing the first satellite image and a second satellite capturing the second satellite image. A similarity of at least one of distribution patterns and densities of objects detected in each of the first satellite image and the second satellite image may be calculated to be used for performance evaluation.

At this time, the multi-satellite performance evaluation system 100 may apply a gray scale to the object distribution image and determine whether or not an object exists in each pixel based on the brightness of each pixel to which the gray scale is applied. Accordingly, the multi-satellite performance evaluation system 100 generates a distribution pattern image by setting the value of each pixel to a 1-bit binary number so that the presence or absence of an object in each pixel is distinguished based on the determined presence or absence. can

Accordingly, the multi-satellite performance evaluation system 100 sets the brightness (or color) of a pixel as a pixel in which no object is detected in the gray scale-applied object distribution image when it is white or black, and the brightness of the pixel ( Alternatively, when the color) is a color other than white or black, the object may be set as a detected pixel.

As another example with reference to FIG. 8 , the multi-satellite performance evaluation system 100 sets a pixel value of 0 or 255 as a pixel in which no object is detected in an object distribution image to which a gray scale is applied, and If the value is 1 to 254, it can be set as the pixel where the object is detected.

In this case, the multi-satellite performance evaluation system 100 may set the value of a pixel where no object is detected as a hash value of 0, and set the value of a pixel where an object is detected as a hash value of 1.

Through this, the multi-satellite performance evaluation system 100 may generate a distribution pattern image 70 in which each of a plurality of pixels is expressed in one of two colors according to whether an object is detected in the object distribution image.

As another example with reference to FIG. 9 , the multi-satellite performance evaluation system 100 uses one (71) of a plurality of pixels included in the distribution pattern image (70) in which the value of each pixel is set to 1 bit according to object detection. ), if there is a pixel 72 where an object is detected among adjacent pixels, one pixel 71 set as a reference may be reset to the same value 73 as the pixel where the object is detected.

At this time, the pixels adjacent to one of the plurality of pixels 71 included in the distribution pattern image 70 are within a preset range (eg, 3 X 3 range or 5 ranges) based on one pixel 71. X 5 range) may be a plurality of pixels included in.

Through the configuration as described above, the multi-satellite performance evaluation system 100 according to the present invention moves the object cluster according to the time difference in the satellite images taken at a predetermined time difference through different satellites 1. can produce a compensating effect.

Furthermore, the multi-satellite performance evaluation system 100 may compare a plurality of distribution pattern images to calculate a similarity of distribution patterns.

Specifically, the multi-satellite performance evaluation system 100 may arrange each of a plurality of distribution pattern images in a binary array according to the order of pixels, and compare whether or not corresponding pixels match each other to calculate a degree of similarity.

Referring to FIG. 10, for example, the multi-satellite performance evaluation system 100 arranges a plurality of pixels included in the first distribution pattern image in a one-dimensional array to generate a first binary array 75, and generates a second distribution pattern image. A second binary array 76 may be generated by arranging a plurality of pixels included in the pattern image in a one-dimensional array.

Accordingly, the multi-satellite performance evaluation system 100 may compare the first binary array 75 and the second binary array 76 to calculate the similarity of the distribution pattern. At this time, the multi-satellite performance evaluation system 100 may check whether corresponding values 78 match at the same position of the first binary array 75 and the second binary array 76 .

To this end, the multi-satellite performance evaluation system 100 may use a Hamming Distance technique for calculating how many different values are present at positions corresponding to each other on data having the same length.

Through this, the multi-satellite performance evaluation system 100 determines the distribution pattern according to the ratio of the total number of pixels included in the distribution pattern image to the number of pixels having identical values in the plurality of distribution pattern images arranged in a binary array. similarity can be calculated.

Therefore, the multi-satellite performance evaluation system 100 may calculate the similarity of the distribution pattern through Equation 1 below.

Meanwhile, the multi-satellite performance evaluation system 100 may apply a gray scale to the object distribution image and generate a histogram based on the brightness of each pixel to which the gray scale is applied.

Referring to FIG. 11, for example, the multi-satellite performance evaluation system 100 calculates the number of pixels for each brightness in the object distribution image to which the gray scale is applied, and creates a graph of the calculated number of pixels for each brightness to form a satellite image. A histogram 80 (Histogram) may be created to indicate the density of objects detected in .

As another example, the multi-satellite performance evaluation system 100 applies a gray scale to each of the first entity distribution image and the second entity distribution image, and the first entity distribution image to which the gray scale is applied and the second entity to which the gray scale is applied Pixel values appearing in each distribution image may be corrected based on preset conditions.

In this case, a preset condition may be set to replace a first pixel value (eg, 255) appearing in each entity distribution image with a second pixel value (eg, 0).

Accordingly, the multi-satellite performance evaluation system 100 determines the number of pixels having the same pixel value size for each pixel value appearing in each of the first entity distribution image whose pixel value is corrected and the second entity distribution image whose pixel value is corrected. A first histogram is generated to indicate the density of objects detected in the first satellite image by calculating the number of pixels for each calculated pixel value as a graph, and a first histogram is generated to indicate the density of objects detected in the second satellite image. 2 Histograms can be created.

Furthermore, the multi-satellite performance evaluation system 100 may compare histograms generated for each of a plurality of object distribution images to calculate a similarity of density.

Specifically, the multi-satellite performance evaluation system 100 may compare the difference between the number of pixels for each pixel value appearing from each of a plurality of histograms to calculate the similarity of density.

For example, the multi-satellite performance evaluation system 100 may calculate a Bhattachayya distance between the first histogram and the second histogram, and determine similarity with respect to density based on the calculated result.

Here, the batachaya distance can be calculated through Equation 2 below, and the similarity to the density according to this can be calculated through Equation 3 below.

은 제 1 히스토그램이고,

는 제 2 히스토그램이며,

은 제 1 히스토그램 및 제 2 히스토그램 각각에 포함된 요소의 개수이며,

는 제 1 히스토그램 및 제 2 히스토그램 각각에 포함된 요소의 인덱스이고,

는 바타차야 거리의 산출 결과에 근거한 고유 값으로서, 값이 0인 경우에 유사도가 100 퍼센트이고, 값이 1인 경우에 유사도가 0 퍼센트일 수 있다.In Equation 2,

is the first histogram,

is the second histogram,

Is the number of elements included in each of the first histogram and the second histogram,

Is an index of an element included in each of the first histogram and the second histogram,

Is a unique value based on the calculation result of the Bhattachaya distance. When the value is 0, the similarity may be 100 percent, and when the value is 1, the similarity may be 0 percent.

As another example, the multi-satellite performance evaluation system 100 calculates a correlation between the first histogram and the second histogram (eg, OpenCV's cv2.HISTCMP_CORREL command can be used), based on the calculated result. The degree of similarity for density can be determined. In this case, the correlation between the first histogram and the second histogram is such that when the calculation result is 1, the first histogram and the second histogram coincide, and when the calculation result is -1, the first histogram and the second histogram are identical. If they are inconsistent and the calculation result is 0, the first histogram and the second histogram may be irrelevant.

As another example, the multi-satellite performance evaluation system 100 performs a Chi-Squared Test between the first histogram and the second histogram (eg, OpenCV's cv2.HISTCMP_CHISQR command may be used). It is calculated, and based on the calculation result, the degree of similarity for density may be determined. In this case, the correlation between the first histogram and the second histogram is consistent with the first histogram and the second histogram when the calculation result is 0, and the first histogram and the second histogram do not match as the value of the calculation result increases. it may be

As another example, the multi-satellite performance evaluation system 100 calculates an intersection relationship between the first histogram and the second histogram (eg, OpenCV's cv2.HISTCMP_INTERSECT command can be used), based on the calculated result. It is possible to determine the degree of similarity for density. In this case, the correlation between the first histogram and the second histogram is consistent with the first histogram and the second histogram when the calculated result is 1, and the first histogram and the second histogram when the calculated result value is 0. This may be inconsistent. In this case, the first histogram and the second histogram may be normalized to a value of 1.

As another example, the multi-satellite performance evaluation system 100 may determine the similarity of the density between the first histogram and the second histogram by using an Earth Mover Distance (EMD) function.

Further, in addition to the methods described above, various methods may be utilized.

Through the configurations described above, the multi-satellite performance evaluation system 100 according to the present invention detects each object from a plurality of satellite images and calculates the similarity of the distribution pattern and density of the object based on the detection result, It can be used to identify performance differences between

In addition, the multi-satellite performance evaluation system 100 calculates the similarity of the distribution pattern and density of objects to the satellite image captured by another satellite based on the satellite image captured by one satellite, thereby calculating the performance of the satellite. It can be used to check the level.

That is, the multi-satellite performance evaluation system 100 calculates the similarity of the distribution pattern and density of the plurality of satellite images captured by the plurality of satellites, respectively, based on the first satellite image, and It can be used to compare liver performance.

In addition, the multi-satellite performance evaluation system 100 is used to check the performance of an object detection algorithm by detecting the same object in satellite images using different methods and calculating the similarity of the object's distribution pattern and density. It could be.

Furthermore, the present invention described above can be implemented as computer readable codes or instructions in a medium on which a program is recorded. That is, various control methods according to the present invention may be integrated or individually provided in the form of a program.

On the other hand, the computer-readable medium includes all types of recording devices in which data readable by a computer system is stored. Examples of computer-readable media include Hard Disk Drive (HDD), Solid State Disk (SSD), Silicon Disk Drive (SDD), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc. there is

Furthermore, the computer-readable medium may be a server or cloud storage that includes storage and can be accessed by electronic devices through communication. In this case, the computer may download the program according to the present invention from a server or cloud storage through wired or wireless communication.

Furthermore, in the present invention, the above-described computer is an electronic device equipped with a processor, that is, a CPU (Central Processing Unit), and there is no particular limitation on its type.

On the other hand, the above detailed description should not be construed as limiting in all respects and should be considered as illustrative. The scope of the present invention should be determined by reasonable interpretation of the appended claims, and all changes within the equivalent scope of the present invention are included in the scope of the present invention.

1: Satellite
10: 1st satellite image
20: second satellite image
30: satellite image
31, 33: interpolation range
40, 50: interpolated satellite image
60: object distribution image
70: distribution pattern image
80: histogram
100: multi-satellite performance evaluation system

Claims (10)

In the multi-satellite performance evaluation method using the multi-satellite performance evaluation system,
inputting a first satellite image and a second satellite image different from the first satellite image;
performing pre-processing on at least one of the first satellite image and the second satellite image, and specifying at least a portion of an area in each of the first satellite image and the second satellite image so that a photographing range overlaps with each other;
A first object distribution image is extracted to reveal a location where an object is detected from a specified area of the first satellite image by using a previously prepared object detection algorithm, and a specified object of the second satellite image is extracted using the object detection algorithm. extracting a second object distribution image to reveal a location where the object is detected from the region;
The first entity distribution image and the second entity distribution image are compared to be used for performance evaluation between the first satellite capturing the first satellite image and the second satellite capturing the second satellite image. calculating a similarity of at least one of distribution patterns and densities of objects detected in each of the first satellite image and the second satellite image; and
and evaluating the performance of at least one of the first satellite and the second satellite by determining a performance difference between the first satellite and the second satellite based on the calculated similarity. .
The method of claim 1 , wherein specifying at least a portion of an area in each of the first satellite image and the second satellite image comprises:
converting each of the first satellite image and the second satellite image into a preset coordinate system so that a comparison of a capturing range between the first satellite image and the second satellite image is possible;
extracting a partial area from one of the converted first satellite image and the converted second satellite image having a larger capturing range so as to overlap a capturing range of another satellite image; and
and reducing a larger one of the first satellite image and the second satellite image having overlapped photographing ranges to correspond to the size of the other satellite image.
The method of claim 1, wherein the extracting of the object distribution image comprises:
correcting molecular scattering reflectance in a specified area of the first satellite image and a specified area of the second satellite image, respectively;
applying a preset mask to remove pixels related to clouds and the land from a specified area of the first satellite image and a specified area of the second satellite image, respectively;
Based on a condition set in advance to extract a pixel likely to be determined as an object, an expected pixel is selected from among a plurality of pixels included in a specified area of the first satellite image and a specified area of the second satellite image, respectively. extracting each;
Calculating a background reflectance of a seawater pixel in a specified area of the first satellite image and a specified area of the second satellite image;
calculating spatial anomaly reflectivity by removing the background reflectivity from the reflectivity of each of a plurality of pixels included in each of the specified area of the first satellite image and the specified area of the second satellite image; and
Among a plurality of pixels included in each of the specified area of the first satellite image and the specified area of the second satellite image, a pixel having a spatial anomaly reflectivity greater than a preset spatial anomaly reflectivity in the noise and visible light bands is selected. A multi-satellite performance evaluation method comprising: extracting as an object;
The method of claim 1, wherein calculating a similarity for at least one of the distribution pattern and density comprises:
applying a gray scale to the first entity distribution image and the second entity distribution image;
determining whether an object exists in each pixel based on brightness of a plurality of pixels included in each of the first object distribution image to which the gray scale is applied and the second object distribution image to which the gray scale is applied; and
Generating a first distribution pattern image and a second distribution pattern image by setting each pixel value to 1 bit so that whether the entity exists in each pixel is distinguished according to whether the determined entity exists or not. A multi-satellite performance evaluation method, including.
5. The method of claim 4, wherein setting each pixel value to 1 bit comprises:
Based on any one of the plurality of pixels included in at least one of the first distribution pattern image and the second distribution pattern image, if a pixel in which an object is detected exists among adjacent pixels, one set as the reference A multi-satellite performance evaluation method that includes;
According to claim 4,
generating a first binary array by arranging a plurality of pixels included in the first distribution pattern image in a one-dimensional array;
generating a second binary array by arranging a plurality of pixels included in the second distribution pattern image in a one-dimensional array; and
Comparing the first binary array and the second binary array to calculate a similarity of the distribution pattern; further comprising, the multi-satellite performance evaluation method.
The method of claim 1, wherein calculating a similarity for at least one of the distribution pattern and density comprises:
applying a gray scale to each of the first entity distribution image and the second entity distribution image;
correcting pixel values appearing in each of the first entity distribution image to which the gray scale is applied and the second entity distribution image to which the gray scale is applied, based on a preset condition;
Calculating the number of pixels having the same pixel value for each pixel value appearing in the first entity distribution image whose pixel values are corrected and the second entity distribution image whose pixel values are corrected; and
A first histogram is created to represent the density of objects detected in the first satellite image by graphing the number of pixels for each of the calculated pixel values, and a first histogram is generated to indicate the density of objects detected in the second satellite image. 2 Step of generating a histogram; including, multi-satellite performance evaluation method.
According to claim 7,
Calculating a difference between the first histogram and the second histogram, and determining a similarity with respect to the density based on the calculated result; further comprising, the multi-satellite performance evaluation method.
a communication unit to which a first satellite image and a second satellite image different from the first satellite image are input; and
Preprocessing is performed on at least one of the first satellite image and the second satellite image to specify at least a portion of an area in each of the first satellite image and the second satellite image so that a photographing range overlaps, and a pre-prepared object A first object distribution image is extracted to reveal a location where an object is detected from a specified area of the first satellite image using a detection algorithm, and an object is extracted from a specified area of the second satellite image using the object detection algorithm. A second entity distribution image is extracted to reveal the detected position, the first entity distribution image and the second entity distribution image are compared, and the first satellite capturing the first satellite image and the second entity distribution image are compared. A similarity of at least one of a distribution pattern and a density of objects detected in each of the first satellite image and the second satellite image is calculated to be used for performance evaluation between the second satellites that have captured the image, and the calculated similarity and a control unit that checks a performance level of at least one of the first satellite and the second satellite by checking a difference in performance between the first satellite and the second satellite based thereon.
A program that is executed by one or more processes in an electronic device and stored in a computer-readable recording medium,
said program,
inputting a first satellite image and a second satellite image different from the first satellite image;
performing pre-processing on at least one of the first satellite image and the second satellite image, and specifying at least a portion of an area in each of the first satellite image and the second satellite image so that a photographing range overlaps;
A first object distribution image is extracted to reveal a location where an object is detected from a specified area of the first satellite image by using a previously prepared object detection algorithm, and a specified object of the second satellite image is extracted using the object detection algorithm. extracting a second object distribution image to reveal a location where the object is detected from the region;
The first entity distribution image and the second entity distribution image are compared to be used for performance evaluation between the first satellite capturing the first satellite image and the second satellite capturing the second satellite image. calculating a similarity of at least one of distribution patterns and densities of objects detected in each of the first satellite image and the second satellite image; and
Including instructions for performing a step of checking a performance level of at least one of the first satellite and the second satellite by checking a performance difference between the first satellite and the second satellite based on the calculated similarity A program stored on a computer-readable recording medium, characterized in that.