How to align satellite images

Abstract

To reduce a positional deviation and a distortion of a satellite image compared with conventional positioning using an SAR image.SOLUTION: A positioning method of a satellite image according to the present invention uses a satellite image based on multi-polarization observation data observed with multi-polarization as a master image 310, acquires information about double scattering intensity, volume scattering intensity and helix scattering intensity using a covariance matrix on the basis of each component about transmission / reception of vertical polarization and horizontal polarization obtained from the multi-polarization observation data, extracts a first feature point whose double scattering intensity is higher than the prescribed intensity and a second feature point whose helix scattering intensity is higher than the prescribed intensity in the master image 310, selects a similar point whose scattered wave intensity is higher than that in the periphery in each slave image 330, and performs positioning of the plurality of slave images 330 to the master image 310 on the basis of a transfer function between each coordinate of the first feature point and second feature point and each coordinate of the similar point.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to a method for aligning satellite images. [Background technology]

[0002] In recent years, Synthetic Aperture Radar (SAR) has been used to observe the state and changes of the Earth's surface over a wide area and with high resolution. With SAR, a satellite equipped with a radar (hereafter referred to as a SAR satellite) or an aircraft is used to irradiate microwaves onto the Earth's surface and receive the waves reflected from the surface, and then combine multiple pieces of information.

[0003] SAR observation data suffers from positional deviations and distortions relative to the target's surface due to deviations in the actual orbit of a satellite or aircraft from a specified orbit, and the effects of weather and water vapor near the target's surface. For example, when using known differential interferometry or time-series interferometry to estimate the location and amount of displacement occurring on the ground, it is necessary to correct positional deviations and distortions in SAR images and align multiple SAR images.

[0004] To align multiple SAR images, it is necessary to extract common feature points (Ground Control Points; GCPs, also called corresponding points) from each SAR image. In addition, it is possible to align the master and slave images by determining the coordinate transformation formula between each of the multiple GCPs selected in one SAR image (master image) and the corresponding GCPs detected in the remaining SAR images (slave images) and applying the determined coordinate transformation formula to the slave images.

[0005] Various methods for aligning multiple SAR images have been proposed, including, for example, a method of correcting the positional deviation between SAR images by using digital elevation model (Digital Elevation Model) of the observed area. Other proposed methods include a method of extracting multiple GCPs common to two or more SAR images obtained by observation, finding their normalized cross-relation functions, and applying the found normalized cross-relation functions to the SAR images (see, for example, Non-Patent Document 1), a method of searching for corresponding points between a master image and a slave image by a phase-only correlation method, aligning the images by robust estimation, calculating a displacement vector, and correcting the positional deviation between the SAR images. [Prior art documents] [Non-patent literature]

[0006] [Non-Patent Document 1] Mikio Tobita, Satoshi Fujiwara, Ryo Murakami, Hiroyuki Nakagawa: High-precision image matching for interferometric SAR, Journal of Geodetic Society, Vol. 45, No. 4, pp. 297-314 (1999). Summary of the Invention [Problem to be solved by the invention]

[0007] When aligning two or more SAR images (satellite images) with high accuracy, it is necessary to extract as many GCPs as possible that are common to the multiple SAR images in each SAR image, with minimal local variation, and on average. In addition, to actually extract GCPs from multiple SAR images, multiple points with high scattered wave strength are selected in one master image, multiple points similar to those points (i.e., points with high scattered wave strength) are detected, and a coordinate transformation formula is obtained using these points as GCPs, and the coordinate transformation value is applied to each slave image. In other words, the accuracy of aligning the master image and slave image is directly affected by the number and relative placement of GCPs.

[0008] However, because the observation conditions are different between the master image and the slave image, even if multiple GCPs with high scattered wave intensity were simply selected in the master image as described above, it was unclear whether those GCPs could be referenced in the slave image, and whether the multiple GCPs selected in the master image were points that did not change over time. Therefore, in conventional registration using SAR images, due to the aforementioned uncertainty, there remained a considerable amount of positional deviation and distortion in the satellite images obtained by SAR observation.

[0009] In addition, it is appropriate to set GCPs for, for example, large structures, long bridges, corners of reclaimed land revetments, intersections, corners of river banks, and bends in roads, but the shape of structures on the ground cannot always be determined by the level of scattered wave intensity in SAR images. Therefore, it is difficult to efficiently extract GCPs suitable for the shapes of the above-mentioned structures on the ground only from the difference in scattered wave intensity in the master image. In that case, it becomes necessary to refer to optical images, maps, etc. acquired separately by optical satellites other than the SAR satellite. In other words, in the past, without combining or referring to optical images, maps, etc., there was a risk that the satellite images of characteristic structures on the ground would be misaligned or distorted to a certain extent in the positioning using conventional SAR images.

[0010] The present invention provides a method for registering satellite images that reduces the misalignment and distortion of satellite images compared to conventional registration using SAR images. [Means for solving the problem]

[0011] The satellite image registration method according to the present invention includes a polarization-specific information acquisition step of acquiring, from the polarization-specific observation data, a VV component representing observation data transmitted with vertical polarization and received with vertical polarization, a VH component representing observation data transmitted with vertical polarization and received with horizontal polarization, an HH component representing observation data transmitted with horizontal polarization and received with horizontal polarization, and an HV component representing observation data transmitted with horizontal polarization and received with vertical polarization, and a covariance matrix of a Pd component representing double scattering intensity, a volume scattering intensity Pd component representing volume scattering intensity, and a volume scattering intensity Pd component representing volume scattering intensity, from information on the four components, the VV component, the VH component, the HH component, and the HV component, the scattering intensity information acquisition step of acquiring information on at least three components, namely, a Pv component representing the intensity of double scattering, a Pc component representing the helix scattering intensity, and a Pc component representing the helix scattering intensity; a feature point extraction step of extracting points in the master image where the double scattering intensity is higher than a predetermined intensity as first feature points and extracting points in the master image where the helix scattering intensity is higher than a predetermined intensity as second feature points; a similarity selection step of using a satellite image to be aligned as a slave image and selecting similar points in the slave image which have a higher scattered wave intensity than the surroundings; and a positioning step of calculating a conversion function between the coordinates of the first feature points and the coordinates of each of the second feature points and the coordinates of the similar points, and positioning the slave image with respect to the master image based on the conversion function.

[0012] In the above-mentioned satellite image registration method, stable feature points (first feature point, second feature point) that are absolute and have little fluctuation over time due to multiple polarization information are extracted from a master image based on multi-polarized observation data. In addition, feature points based on information on structures on the ground surface, which are difficult to extract from single-polarized observation data alone, are extracted from the master image. Therefore, according to the above-mentioned satellite image registration method, the positions of all slave images to be registered can be registered with high accuracy based on the master image based on multi-polarized observation data. Therefore, by introducing a master image based on multi-polarized observation data, it is possible to reduce positional deviation and distortion of the slave images to be registered compared to conventional registration using SAR images based only on single-polarized observation data.

[0013] Furthermore, in the satellite image registration method according to the present invention, when at least one of the total number of the first feature points and the second feature points is less than a predetermined number and the first feature points and the second feature points are unevenly distributed in the feature point extraction step, a point in the master image where the volume scattering intensity is higher than a predetermined intensity may be extracted as a third feature point, and in the registration step, a conversion function between the coordinates of the first feature point, the second feature point, and the third feature point and the coordinates of the similar point may be calculated, and the slave image may be registered with the master image based on the conversion function.

[0014] In the above-described method for aligning satellite images, the predetermined number and the degree of uneven distribution of feature points are appropriately set, thereby further improving the accuracy of alignment between the master image and each of the multiple slave images compared to conventional alignment using SAR images based only on single-polarized observation data.

[0015] Furthermore, in the satellite image alignment method according to the present invention, in the scattering intensity information acquisition step, a Ps component representing surface scattering intensity may be acquired from information regarding the four components using a covariance matrix, and the Ps component, Pd component, and Pv component may be synthesized to calculate the level of the helix scattering intensity in the master image to acquire information regarding the Pc component.

[0016] In the above-mentioned method for aligning satellite images, for example, the Ps, Pd, and Pv components are treated as image information quantitatively expressed by the gradation values ​​of the three primary colors of light, and by synthesizing RGB images, it is possible to easily grasp information on the Pc component from the brightness and color balance of white in the synthesized image. Therefore, information on the Pc component related to the helix scattering intensity in the master image can be easily obtained. Effect of the Invention

[0017] According to the present invention, it is possible to reduce positional deviation and distortion of satellite images compared to conventional position alignment using SAR images. [Brief description of the drawings]

[0018] [Figure 1] 1 is a schematic diagram of a satellite image correction system and associated components for implementing a satellite image alignment method according to one embodiment of the present invention; [Diagram 2] 4 is a flowchart showing the process of a satellite image alignment method according to an embodiment of the present invention. [Diagram 3] FIG. 1 is a conceptual diagram for explaining a method for aligning satellite images according to an embodiment of the present invention. [Figure 4] FIG. 1 is a conceptual diagram for explaining a method for aligning satellite images according to an embodiment of the present invention. [Diagram 5] FIG. 1 is a conceptual diagram for explaining a conventional method for aligning satellite images. [Figure 6] FIG. 1 is a conceptual diagram for explaining a conventional method for aligning satellite images. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0019] Hereinafter, a method for aligning satellite images according to the present invention will be described with reference to the drawings.

[0020] 1, an earth observation system 50 observes the earth's surface G (i.e., the earth's surface) using a satellite image alignment method according to an embodiment of the present invention. The earth observation system 50 includes a plurality of SAR satellites (artificial satellites) 100 and an alignment system 200 capable of communicating with each of the plurality of SAR satellites 100.

[0021] Each of the multiple SAR satellites 100 revolves around a predetermined orbit above the earth's surface G. FIG. 1 shows two of the multiple SAR satellites 100, SAR satellites 101 and 102. At least the SAR satellite 100 of the multiple SAR satellites 100 is equipped with a SAR (synthetic aperture laser) and is configured to be able to observe an observation region R of the earth's surface G with multiple polarizations. The SAR satellite 100 is equipped with, for example, an antenna capable of transmitting and receiving H polarization (horizontal polarization) horizontal to the earth's surface G, an antenna capable of transmitting and receiving V polarization (vertical polarization) perpendicular to the earth's surface G, and a signal processing unit configured to be able to input and output radio signals related to the H polarization and V polarization to and from each of the antennas. Of the multiple SAR satellites 100, the SAR satellite 102 is configured to be able to observe the observation region R with a single polarization.

[0022] The positioning system 200 is installed in a base station or the like provided on the earth's surface G, and is composed of a calculator or personal computer or the like having a built-in program that performs the processing described below. In addition to the positioning system 200, the above-mentioned base station is also provided with a control system that controls the multiple SAR satellites 100 and adjusts the orbits of their circular movement, a power source for operating each system, and an antenna for transmitting and receiving various signals to and from each of the multiple SAR satellites 100. The positioning system 200 is capable of wireless communication with each of the multiple SAR satellites 100.

[0023] 2 is a flowchart of a satellite image registration method according to an embodiment of the present invention, which is performed by the registration system 200. The satellite image registration method according to this embodiment includes at least a polarization-specific information acquisition step, a scattering intensity information acquisition step, a feature point extraction step, and a registration step. Each step will be described below.

[0024] <Polarization-specific information acquisition process> In the polarization-specific information acquisition step, the alignment system 200 receives polarized observation data 151 observed with multiple polarizations from the SAR satellite 101, and creates a polarized SAR image (satellite image observed with multiple polarizations) based on the polarized observation data 151 (step S251). The polarized observation data 151 includes, for example, the direction of polarization (V / H), the wavelength of the radio waves, and the angle of the path of the radio waves emitted from the SAR satellite 101 with respect to the Earth's surface G. The alignment system 200 sets the created polarized SAR image as the master image 310 (see FIG. 3) to be aligned.

[0025] The alignment system 200 breaks down the polarized observation data 151 into four components, a VV component, a VH component, an HH component, and an HV component, and acquires information on each component (step S252). The VV component represents observation data transmitted to the observation region R with V polarization and received from the observation region R with V polarization. The VH component represents observation data transmitted to the observation region R with V polarization and received from the observation region R with H polarization. The HH component represents observation data transmitted to the observation region R with H polarization and received from the observation region R with H polarization. The HV component represents observation data transmitted to the observation region R with H polarization and received from the observation region R with H polarization.

[0026] <Scattering intensity information acquisition process> In the scattering intensity information acquisition process, the alignment system 200 calculates at least three components, namely, the Pd component, the Pv component, and the Pc component, from the four components, namely, the VV component, the VH component, the HH component, and the HV component, and their information, using a covariance matrix (so-called coherency matrix), and acquires information about these three components (step S253).

[0027] The Pd component represents the two-time scattering intensity, and is mainly generated from, for example, the perpendicular structures between the ground or sea surface of the earth's surface G and the sides of man-made objects or tree trunks. The Pv component represents the volume scattering intensity, and is mainly generated from, for example, vegetation on the earth's surface G. The Pc component represents the helix scattering intensity, and changes linear polarization to circular polarization. For example, by applying a four-component decomposition algorithm based on the covariance matrix to the above-mentioned four components of multi-polarized waves, at least the above-mentioned three components can be calculated, and the Ps component can also be calculated. The Ps component represents the surface scattering intensity, and is mainly generated from, for example, the flat ground or sea surface of the earth's surface G.

[0028] <Feature point extraction process> 3, in the feature point extraction step, the registration system 200 extracts points in the master image 310 where the second scattering intensity (i.e., the Pd component) is higher than a predetermined intensity as first feature points GCP1 (step S254), and extracts points in the master image 310 where the helix scattering intensity (i.e., the Pc component) is higher than a predetermined intensity as second feature points GCP2 (step S255). For example, in a master image (not shown) in which the level of second scattering intensity is quantitatively represented by the gradation value of red (R) out of the three primary colors of light, the coordinates (x 1-1 ,y 1-1 ), (x 1-2 ,y 1-2 ), …, (x 1-a ,y 1-a ) and reflecting them in the master image 310, the first feature points GCP1-1, ..., GCP1-a can be extracted. The predetermined intensity for the two-time scattering intensity can be set appropriately. a is a natural number that changes depending on the predetermined intensity for the two-time scattering intensity.

[0029] In a master image (not shown) in which the volume scattering intensity (i.e., the Pv component) is quantitatively expressed by the gradation value of green (G) out of the three primary colors of light, the coordinates (x 3-1 ,y 3-1 ), (x 3-2 ,y 3-2 ), …, (x 3-b ,y3-b ) and reflecting them in the master image 310, third feature points GCP3-1, ..., GCP3-b can be extracted. The predetermined intensity for the volume scattering intensity can be set appropriately. b is a natural number that changes depending on the predetermined intensity for the volume scattering intensity. The level of the surface scattering intensity (i.e., the Ps component) is quantitatively expressed and imaged using the gradation value of blue (B) out of the three primary colors of light. The Ps component, which roughly corresponds to the scattered wave intensity in the SAR image in the conventional single-polarized observation, is relatively low, and at least lower than the threshold value corresponding to the predetermined intensity for the surface scattering intensity, unless the observation target on the earth's surface G is directly facing the SAR satellite.

[0030] The level of the Pc component of the helix scattering intensity can be determined based on the composite value of the Pd, Pv, and Ps components. Specifically, the three images of the red image related to the Pd component, the green image related to the Pv component, and the blue image related to the Ps component are composited, and the coordinates (x,y,z) of the points where the helix scattering intensity, expressed as the composite color of the three primary colors, i.e., the brightness of white, is higher than a predetermined intensity, are calculated. 2-1 ,y 2-1 ), (x 2-2 ,y 2-2 ), …, (x 2-d ,y 2-d ) and reflecting them in the master image 310, they can be extracted as second feature points GCP2-1, ..., GCP2-d. The predetermined intensity related to the helix scattering intensity can be set appropriately. d is a natural number that changes depending on the predetermined intensity related to the helix intensity. In this specification, when describing the contents common to any of the first, second, and third feature points GCP#-1, ..., GCP#-(any of a, d, and b), these feature points will be collectively referred to as GCP#.

[0031] As feature points in the master image 310, a first feature point GCP1 resulting from the Pd component and a second feature point GCP2 resulting from the Pc component are extracted. However, if the sum (a+d) of the total number a of the first feature points GCP1 and the total number d of the second feature points GCP2 (total number of the first feature points and the second feature points) is less than a predetermined number, it is preferable to extract a third feature point GCP3 in addition to the first feature point GCP1 and the second feature point GCP2 as a feature point in the master image 310 (step S256). The predetermined number is set appropriately taking into account the environment and conditions when the observation data is acquired.

[0032] Also, when at least one of the first feature point GCP1 and the second feature point GCP2 is unevenly distributed in the master image 310, it is preferable to extract the third feature point GCP3 as a feature point in the master image 310 in addition to the first feature point GCP1 and the second feature point GCP2 (step S257). The uneven distribution degrees that serve as thresholds for each of the first feature point GCP1 and the second feature point GCP2 are appropriately set in consideration of the environment, conditions, etc. at the time of acquiring the observation data.

[0033] <Similar point selection process> In the similarity selection step, the alignment system 200 receives single-polarized observation data 152 observed with single polarization from the SAR satellite 102 (or a SAR satellite among the multiple SAR satellites 100 that performs single-polarized observation other than the SAR satellite 102), and creates a single-polarized SAR image based on the single-polarized observation data 152 (step S258). The single-polarized observation data 152 includes, for example, the direction of polarization (V / H), the wavelength of the radio waves, and the angle of the path of the radio waves emitted from the SAR satellite 102 with respect to the Earth's surface G. The alignment system 200 uses the created single-polarized SAR image as slave images 331, 332 (see FIG. 3) to be aligned. Note that, although FIG. 3 illustrates only two slave images 331, 332, the number (total number) of slave images may be three or more and is not limited to a specific number. Hereinafter, when describing common content among multiple slave images, these slave images will be collectively referred to as slave image 330. The total number of slave images 330 is, for example, 10 or more, and preferably 20 or more.

[0034] In each of the slave images 330, points that look similar even if the scattering intensity is not necessarily high, i.e., points with a higher scattering wave intensity than the surroundings, are selected as similar points AP (step S259). For example, as shown in FIG. 3, in the slave image 331, a plurality of similar points AP with a higher scattering wave intensity than the surroundings are selected as similar points AP. 1-1 , AP 1-2 , … , AP 1-e and calculate the coordinates of each similarity point (x´ 1-1 ,y´ 1-1 ), (x´ 1-2 ,y´ 1-2 ), …, (x´ 1-e ,y´ 1-e ) is acquired. In addition, multiple similar points AP 2-1 , AP 2-2 , … , AP 2-e and calculate the coordinates of each similarity point (x´ 2-1 ,y´ 2-1 ), (x´ 2-2 ,y´ 2-2 ), …, (x´ 2-e ,y´ 2-e ) is obtained. e is a natural number that is the same for the multiple slave images 330, and is preferably the same as the total number of the first feature points GCP1, the second feature points GCP2, and the third feature points GCP3, if necessary, in the master image 310. In this specification, the similar points AP selected in each of the multiple slave images 330 are referred to as 1-1 , AP 1-2 , … , AP 1-e When describing the commonalities between the two, these similarities will be collectively referred to as AP.

[0035] <Alignment process> In the registration process, the registration system 200 calculates a conversion function f between the coordinates (x, y) of the first feature point GCP1, the second feature point GCP2, and, if necessary, the third feature point GCP3 and the coordinates (x', y') of the similarity point AP. n(Step S260), where n represents the number of the slave images 330. That is, the transformation function of the slave image 331 with respect to the master image 310 is represented as f1. As shown in FIG. 3, for example, the transformation function f1 is expressed as 1-1 ,y 1-1 ) and similarities AP 1-1 Coordinates (x´ 1-1 ,y´ 1-1 ), …, the coordinates of the second feature point GCP2-1 (x 2-1 ,y 2-1 ) and similarities AP 1-3 Coordinates (x´ 1-3 ,y´ 1-3 ) …, the coordinates of the third feature point GCP3-1 (x 3-1 ,y 3-1 ) and similarities AP 1-5 Coordinates (x´ 1-5 ,y´ 1-5 Similarly, the transformation function f2 of the slave image 332 with respect to the master image 310 is calculated as a transformation variable between the coordinates (x 1-1 ,y 1-1 ) and similarities AP 2-1 Coordinates (x´ 2-1 ,y´ 2-1 ), …, the coordinates of the second feature point GCP2-1 (x 2-1 ,y 2-1 ) and similarities AP 2-3 Coordinates (x´ 2-3 ,y´ 2-3 ) …, the coordinates of the third feature point GCP3-1 (x 3-1 ,y 3-1 ) and similarities AP 2-5 Coordinates (x´ 2-5 ,y´ 2-5 ) are calculated as the respective transformation variables between

[0036] As shown in FIGS. 3 and 4, the alignment system 200 is based on a transformation function f n Based on this, the master image 310 and the n-th slave image 330 are aligned with each other (step S261). By performing the above steps, it is possible to align each of the multiple slave images 330 observed with a single polarization with the master image 310 observed with multiple polarizations.

[0037] The above-described satellite image alignment method of the present embodiment includes a polarization-specific information acquisition step, a scattering intensity information acquisition step, a feature point extraction step, a similar point selection step, and an alignment step. In the polarization-specific information acquisition step, a SAR image (satellite image) based on multi-polarized observation data 151 observed by a SAR satellite (artificial satellite equipped with a synthetic aperture radar) is used as a master image 310 to be aligned, and a VV component representing observation data transmitted with V polarization and received with V polarization, a VH component representing observation data transmitted with V polarization and received with H polarization, an HH component representing observation data transmitted with H polarization and received with H polarization, and an HV component representing observation data transmitted with H polarization and received with V polarization are acquired from the multi-polarized observation data 151. In the scattering intensity information acquisition step, information on at least three components, namely, a Pd component representing the double scattering intensity, a Pv component representing the volume scattering intensity, and a Pc component representing the helix scattering intensity, is acquired from information on the four components, namely, the VV component, the VH component, the HH component, and the HV component, using a covariance matrix. In this embodiment, in addition to the above-mentioned three components, information on the Ps component representing the surface scattering intensity is also acquired. In the feature point extraction step, a point in the master image 310 where the double scattering intensity is higher than a predetermined intensity is extracted as a first feature point GCP1, and a point in the master image 310 where the helix scattering intensity is higher than a predetermined intensity is extracted as a second feature point GCP2. In the similarity point selection step, similar to the conventional satellite image registration, the satellite image to be registered is set as the slave image 330, and a similarity point AP where the scattered wave intensity is higher than the surroundings is selected in each slave image 330. In the registration step, a conversion function f between each coordinate of the first feature point GCP1 or the second feature point GCP2 and each coordinate of the similarity point AP is calculated. n Calculate the conversion function f n The slave images 330 are aligned relative to the master image 310 based on the above.

[0038] According to the satellite image alignment method of this embodiment, since the master image 310 based on the multi-polarized observation data 151 is used as a reference, it is possible to extract stable feature points (first feature point GCP1, second feature point GCP2) that are absolute and have little fluctuation over time based on multiple polarization (V / H) information in the master image 310. In addition, it is possible to extract feature points based on information on, for example, large structures, long bridges, corners of revetments on reclaimed land, intersections, corners of river banks, and road bends, which are difficult to extract from single-polarized observation data alone, in the master image 310. Therefore, according to the satellite image alignment method of this embodiment, it is possible to align the positions of all slave images 330 to be aligned with high accuracy based on the master image 310. On the other hand, in the conventional satellite image alignment method, as shown in FIG. 5, one SAR image out of multiple SAR images based on single-polarized observation data is set as a master image 351, and the remaining SAR images are set as slave images 352, ... (only the slave image 352 is illustrated in FIG. 5). A number of points in the master image 351 that have higher scattered wave intensity than the surroundings are called similar points BP1, ..., BP k (In FIGS. 5 and 6, k=5), and multiple points in each slave image 350 where the scattered wave intensity is higher than the surroundings are designated as similar points AP1, . . . , AP k Let the similarities be BP1, ..., BP k Each coordinate (x,y) of AP1, ..., AP k A transformation function f is calculated between each coordinate (x', y') of the similar points BP1, ..., BP k Similarities to AP1, …, AP k 6, similar points BP and AP are significantly misaligned, resulting in misalignment and distortion. However, by introducing a master image 310 based on multi-polarized observation data 151, it is possible to reduce misalignment and distortion of the slave image (satellite image) 330 to be aligned compared to conventional alignment using SAR images based only on single-polarized observation data.

[0039] In the satellite image registration method of this embodiment, in the feature point extraction step, when the total number (a+d) of the first feature points GCP1 and the second feature points GCP2 is less than a predetermined number, or when the first feature points GCP1 and the second feature points GCP2 are unevenly distributed, a point in the master image 310 whose volume scattering intensity is higher than a predetermined intensity is extracted as the third feature point GCP3. In addition, in the registration step, a transformation function f between each coordinate of the third feature point GCP3 and the coordinate of the similar point is calculated in addition to the coordinates of the first feature point GCP1 and the second feature point GCP2. n Calculate the number and convert it to the conversion function f n The slave images 330 are aligned with the master image 310 based on the above.

[0040] In the satellite image registration method of this embodiment, in order to adjust the number and unevenness of feature points in the master image 310 to a number and unevenness that can improve the accuracy of registration between the master image 310 and each of the multiple slave images 330, a third feature point GCP3 can be introduced in addition to the first feature point GCP1 and the second feature point GCP2. In other words, according to the satellite image registration method of this embodiment, it is possible to further improve the accuracy of registration between the master image 310 and each of the multiple slave images 330 compared to conventional registration using SAR images based only on single-polarized observation data.

[0041] Furthermore, in the satellite image alignment method of this embodiment, in the scattering intensity image acquisition process, the Ps component representing the surface scattering intensity is obtained using a covariance matrix from information on the four components, the VV component, the VH component, the HH component, and the HV component, and the Ps component, Pd component, and Pv component are synthesized to calculate the level of helix scattering intensity in the master image 310, thereby obtaining information on the Pc component.

[0042] According to the satellite image registration method of this embodiment, for example, the Ps, Pd, and Pv components of the polarimetric observation data are expressed quantitatively as gradation values ​​of the three primary colors of light (RGB), and by synthesizing the RGB images, the Pc component information can be easily grasped from the brightness and color balance of white in the synthesized image. Therefore, according to the satellite image registration method of this embodiment, the Pc component information related to the helix scattering intensity in the master image 310 can be easily obtained.

[0043] Although the preferred embodiment of the present invention has been described in detail above, the present invention is not limited to the specific embodiment, and the present invention can be modified within the scope of the spirit of the present invention described in the claims.

[0044] For example, in the above-described embodiment, SAR satellite 101 and SAR satellite 102 may be configured as the same SAR satellite, and the SAR satellite may be configured to be capable of observation with both multi-polarized waves and single-polarized waves. In that case, for example, the multiple slave images 330 may be based on single-polarized data observed when the SAR satellite orbited toward observation region R at a time different from the time when the master image 310 was observed. Also, in the above-described embodiment, the slave image 330 may be one image.

[0045] For example, in the above embodiment, it has been explained that information on the Pc component related to helix scattering intensity can be obtained by treating the Ps component, Pd component, and Pv component of the polarimetric observation data as image information quantitatively represented by the gradation values ​​of the three primary colors of light and synthesizing the RGB images. However, it is also possible to quantitatively confirm the level of the Pc component directly from information on the four components of the VV component, VH component, HH component, and HV component of the polarimetric observation data, based on the four components calculated using a covariance matrix. [Explanation of symbols]

[0046] 100, 101, 102…SAR satellite (artificial satellite) 200…Alignment system

Claims

1. a polarization-specific information acquisition process for acquiring, from the polarization observation data, a VV component representing observation data transmitted with vertical polarization and received with vertical polarization, a VH component representing observation data transmitted with vertical polarization and received with horizontal polarization, an HH component representing observation data transmitted with horizontal polarization and received with horizontal polarization, and an HV component representing observation data transmitted with horizontal polarization and received with vertical polarization, using a satellite image based on polarized observation data observed with multiple polarized waves by an artificial satellite equipped with a synthetic aperture radar as a master image to be aligned; a scattering intensity information acquisition step of acquiring information on at least three components, a Pd component representing two-time scattering intensity, a Pv component representing volume scattering intensity, and a Pc component representing helix scattering intensity, from information on the four components, the VV component, the VH component, the HH component, and the HV component, using a covariance matrix; a feature point extraction step of extracting, as first feature points, points in the master image where the second scattering intensity is higher than a predetermined intensity, and extracting, as second feature points, points in the master image where the helix scattering intensity is higher than a predetermined intensity; a similarity point selection step of selecting a similarity point in the slave image, the similarity point having a scattered wave intensity higher than a predetermined intensity set in advance, compared to the surrounding area, using the satellite image to be aligned as a slave image; a registration process of calculating a conversion function between the coordinates of the first feature point and the coordinates of the second feature point and the coordinates of the similar point, and aligning the slave image with the master image based on the conversion function; Equipped with In the feature point extraction step, when the total number of the first feature points and the second feature points is less than a predetermined number, or when the first feature points and the second feature points are unevenly distributed, In the feature point extraction step, extracting a point in the master image where the volume scattering intensity is higher than a predetermined intensity as a third feature point; In the alignment process, calculating a transformation function between the coordinates of the first feature point, the second feature point, and the third feature point and the coordinates of the similar point, and aligning the slave image with the master image based on the transformation function; Satellite image alignment methods.

2. In the scattering intensity information acquisition step, From the information on the four components, a Ps component representing the surface scattering intensity is obtained using a covariance matrix; the Ps component, the Pd component, and the Pv component are synthesized, and the level of the helix scattering intensity in the master image is calculated to obtain information about the Pc component. The method for aligning satellite images according to claim 1 .

Images