KR100870894B1 - Method of automatic geometric correction for linear pushbroom image - Google Patents

Abstract

An automatic geometric correction method of a linear scanning image is provided to extract a ground reference point automatically from a digital elevation model, thereby assigning an accurate ground coordinate to each image coordinate. A first ground coordinate(P1x,P1y,P1z) is extracted from the corresponding point and auxiliary data(30). A second ground coordinate(P2x,P2y,P2z) is extracted by using a correlation coefficient of the first ground coordinate and a digital elevation model(40). A photographing position(Sx,Sy,Sz) and posture are improved among the auxiliary data by using the second ground coordinate(60). An outer orientation corrects distortion of an image by assigning the ground coordinate to each image coordinate of an image by using the second ground coordinate(70).

Description

Automatic Image Geometry Correction for Linear Scanning Image {METHOD OF AUTOMATIC GEOMETRIC CORRECTION FOR LINEAR PUSHBROOM IMAGE}

1 is a process flow diagram of an automatic geometric correction method according to the present invention.

FIG. 2 is a diagram illustrating an error of first ground coordinates calculated using a SPOT satellite, which is a linear scanning satellite image for 35 corresponding points. FIG.

3 is a flow chart of a second method of extracting the ground coordinates according to the present invention.

4 is a diagram illustrating an example of a correlation coefficient map calculated from a SPOT 3 image, which is a linear scanning image;

5 shows an example of interpolating a position having a maximum value of a correlation coefficient map from a SPOT 3 image, which is a linear scanning image.

FIG. 6 is a graph showing an example of an error amount distribution in the X and Y directions for a SPOT 3 satellite that is a linear scanning satellite image; FIG.

7 is a diagram illustrating an embodiment of a partition determined for correction of an error amount in the X and Y directions.

FIG. 8 is a view showing a difference from second terrestrial ground coordinates determined by using the SPOT 3 satellite, which is a linear scanning satellite image for 35 corresponding points; FIG.

)과 자세오차관측각(

)의 기하를 나타낸 도면,9 is a position error observation angle (

) And posture error

Drawing showing the geometry of

의 기하를 나타낸 도면,10 is a shooting position error

Drawing showing the geometry of

)의 기하를 나타낸 도면이다.11 is a yaw error (

Is a diagram showing the geometry of.

The present invention relates to a method of automatically correcting the geometric distortion of a linear scanning image in order to explore the characteristics of the surface, and more particularly, to the observation position of two or more linear scanning images of the same area photographed at different viewing angles. The present invention relates to a method of automatically extracting a ground reference point from a digital elevation model (DEM) using a speed, an observation angle, and the like to give an accurate ground coordinate to each image coordinate of an image.

에 대응하는 지상점

을 부여한 지 상기준점(GCP)를 이용한다. 수 개 이상의 지상기준점을 이용하여 기하를 복원하고, 상기 복원한 기하로부터 영상에 지상좌표를 부여하여 영상에 정확한 지상위치를 부여하게 된다. In the case of linear scanning images, the observation position and velocity, observation posture, and angle of view given in the supplementary data are generally not accurate. The location can't be determined. This is especially true for satellites. Image Points to Correct Geometric Distortion of Linear Scanned Images

Ground point corresponding to

Use the GCP to give a value. Geometry is restored using several or more ground reference points, and the ground coordinates are given to the image from the restored geometry to give an accurate ground position to the image.

In order to use the ground control point, it is necessary to know the exact ground point and observe the image point corresponding to the ground point. This requires a lot of time and money. Therefore, many methods have been proposed to reduce the number of ground control points.

One method was to reduce the number of ground control points by adjusting only the shooting angle (LOS vector) of the sensor. However, the use of a small amount of ground control points may be distorted due to an error in the selection of image points in the image, a positional error with the ground points, and furthermore, it is impossible to use the above method in an area where ground control points cannot be obtained. there was.

Meanwhile, the United States currently provides Shuttle Radar Topography Mission (SRTM) Digital Elevation Model (DEM). The SRTM DEM provides numerical elevation data with a spatial resolution of 1 second (approximately 30 m) in the United States, and a numerical elevation with a spatial resolution of 3 seconds (approximately 90 m) in the world except the United States (latitude N60 degrees to S56 degrees). Provides data. If the digital elevation data can be used to correct the geometric distortion of the linear scanning image, the time and cost of extracting the ground reference point in the world will be significantly reduced.

The present invention has been made in view of the above, by using the digital elevation data provided from the digital elevation model (DEM) to automatically extract the ground reference point to give the correct ground coordinates to each image coordinate of the image, The purpose of the present invention is to provide an automatic image geometry correction method for linearly scanned images, which saves time and costs for correction and improves the accuracy of terrestrial coordinates.

Automatic image geometry correction method of the linear scanning image of the present invention for achieving the above object,

(a) extracting two or more different images of the same ground surface and auxiliary data for the images;

(b) extracting corresponding points through image matching from two or more different images;

를 추출하는 단계; (c) a first ground coordinate from the correspondence point and the ancillary data;

Extracting;

를 추출하는 단계; (d) the second ground coordinate using the correlation coefficient between the first ground coordinate and the numerical elevation data;

Extracting;

및 자세

를 개선하는 단계; (e) a photographing position of the auxiliary data using the second ground coordinates;

And posture

Improving the;

and (f) performing external coordinates to correct distortion of the image by assigning the ground coordinates to each image coordinate of the image using the second ground coordinates.

In particular, step (d),

A first coordinate conversion step of converting the first ground coordinates into a coordinate system having the numerical elevation data;

Producing a correlation coefficient map using the converted first ground coordinates and an elevation value of the numerical elevation data;

Determining a maximum correlation coefficient position in the produced correlation coefficient map;

Correcting the error amount extracted in the determining of the maximum correlation coefficient position;

And a second coordinate transformation for inversely transforming the coordinate where the error amount is corrected into a coordinate system before transformation.

In addition, the auxiliary data of step (a) may include information about the position of the photographing apparatus, the speed information, the attitude information of the photographing apparatus, the photographing time of the center line of the image, the photographing time per line of the image, the photographing angle, and the like. It features.

In addition, the step (b) is characterized by including the process of finding the highest correlation coefficient by calculating the correlation coefficient for different images.

In step (e),

과

and

를 이용하여 오차관측각

,

을 계산하는 단계;

Error observation angle using

,

Calculating;

The error observation angle

과

and

에 대입하여 촬영위치오차

와 자세오차

를 계산하며, 상기

와

는 요각 오차

의 X와 Y성분,

는 촬영위치와 지상위치의 수직성분으로부터의 각도 (

), Rx, Ry는 R은 촬영위치와 지상위치 사이의 거리의 X, Y 성분인 것을 특징으로 한다.

Position error

And posture error

Calculate the above

Wow

Is the yaw error

X and Y components of,

Is the angle from the vertical component of the shooting position and the ground position (

), Rx, Ry, R is characterized in that the X, Y components of the distance between the shooting position and the ground position.

In addition, the step of producing a correlation coefficient map using the converted first ground coordinates and the altitude value of the numerical elevation data,

According to the accuracy of the auxiliary data extracted in step (a), the minimum and maximum values of the error are set, and the interval of the error is set, and then the correlation coefficient map is produced from the error interval from the minimum value of the error to the maximum value of the error. Characterized in that.

In addition, when the SPOT 3 satellite is used, the minimum error value is -1500m, the maximum error value is 1500m, and the error interval is determined as 10m.

In addition, determining the maximum correlation coefficient position in the produced correlation coefficient map,

Determining a maximum value in the produced correlation coefficient map;

Performing interpolation in sub-pixel units in the X and Y directions from a position where the correlation coefficient is maximum;

Determining a position where a correlation coefficient is maximum by performing the interpolation;

)와 Y방향의 오차(

)를 결정하는 단계;Using this crystal, the error in the X direction (

) And the error in the Y direction (

Determining;

And determining an altitude error using the altitude extracted from the digital elevation data and the altitude of the first ground coordinate using the error.

In addition, the step of correcting the error amount extracted in the step of determining the maximum correlation coefficient position,

The error amount in the X direction, the error in the Y direction, and the altitude error are differently corrected according to the X and Y directions.

In addition, the numerical elevation data is characterized in that provided in the Shuttle Radar Topography Mission (SRTM) Digital Elevation Model (DEM).

In addition, the first coordinate transformation step of converting the first ground coordinates into a coordinate system of the numerical elevation data,

When using the numerical elevation data provided by STRM DEM, it is characterized by converting given coordinates to longitude coordinates.

The terminology used herein is for the purpose of describing particular example embodiments only and is not intended to be limiting of the present invention. A singular expression includes a plural expression unless the context clearly indicates otherwise. In this application, the terms "comprise" or "having" are intended to indicate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, and one or more other It is to be understood that the present invention does not exclude the possibility of the presence or the addition of features, numbers, steps, operations, components, parts, or a combination thereof.

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

1 is a flowchart illustrating a process of performing automatic geometric correction of a linear scanning image according to the present invention.

를 추출하며(30), 상기 제 1 지상좌표와 수치표고자료의 상관계수(50)를 이용하여 제 2 지상좌표

를 추출(40)한다. 이후 상기 제 2 지상좌표를 이용하여 상기 보조자료 중 촬영위치

및 자세

를 개선(60)하고, 상기 제 2 지상좌표를 이용하여 영상의 각 영상좌표에 지상좌표를 부여함으로서 영상의 왜곡을 보정하는 외부표정 을 수행(70)한다.After extracting two or more images of the ground surface and auxiliary data for the images (10), and extracting the corresponding points through image matching from these images (20), the first ground coordinates from the corresponding points and the auxiliary data

(30), the second ground coordinates using the correlation coefficient (50) of the first ground coordinates and the numerical elevation data

Extract 40. Then, the photographing position of the auxiliary data using the second ground coordinates

And posture

Improve (60) and external coordinates to correct distortion of the image by assigning the ground coordinates to each image coordinate of the image using the second ground coordinates (70).

Hereinafter, each step will be described in more detail with reference to the accompanying drawings.

The image and auxiliary data extraction step 10 extracts a stereo image of the same region and extracts auxiliary data for each image.

Here, the stereoscopic image is a pair of images that allows to see the same area in three dimensions by using the super-stereoscopic effect, and the auxiliary data includes the position and speed information of the shooting device, the attitude information of the shooting device, the recording time of the center line of the image, Information on shooting time, shooting angle, etc. per line. In order to perform the automatic geometric correction according to the present invention, it is necessary to extract two or more images photographing the same region and auxiliary data thereof.

의 윈도우 크기가 되도록 영상패치를 제작한 후, 다른 영상의

의 영역에서 영상패치와의 상관계수를 계산한다. 상기에 계산된 상관계수 중 이 값이 가장 높은값을 지니는 위치를 대응점으로 선택한다. 이러한 과정은 수치사진측량 분야에서 일반적으로 사용되는 당업자에게 널리 알려진 과정이다.The corresponding point extracting step 20 is a step of finding identical points corresponding to each other in the input image. The correspondence point is a point representing the same position in different images. Extraction of the corresponding point requires a process of finding the highest correlation coefficient by calculating correlation coefficients for different images. One image is selected as the reference image, and a point where the change in brightness value is relatively large is selected in the selected image. From the selected point

After creating the image patch to be the window size of the

Calculate the correlation coefficient with the image patch in the region of. The position having the highest value among the correlation coefficients calculated above is selected as the corresponding point. This process is well known to those of ordinary skill in the art commonly used in the field of digital photogrammetry.

를 추출하는 단계이다. 제 1지상좌표의 추출은 한국특허 출원번호 제10-2006-26655호 "LOS벡터 조정 모델을 이용한 영상의 기하보정 방법 및 그 장치"의 LOS 벡터 조정 모델을 통하여 이루어진다. The first ground coordinate extraction step 30 is a first ground coordinate from the auxiliary data extracted in the above and the corresponding point extracted in the corresponding point extraction step 20

Extracting step. Extraction of the first ground coordinates is performed through the LOS vector adjustment model of Korean Patent Application No. 10-2006-26655, "Geometry of Image Correction Method and Apparatus Using the LOS Vector Adjustment Model".

The LOS vector adjustment model is represented by Equations 1 and 2 below.

는

과

의 단위를 지상에서의 거리로 환산하기 위하여 사용한 변수이고, 영상점

는 영상좌표로 정의되며, 이 영상점에 대한 지상점

은 지심좌표계(Earth-Centered Earth-Fixed Coordinate System; ECEF)로 정의된다.

와

는 지상점

과 촬영위치

사이의 벡터에 의해서 정의되는 제 1 관측각이며, 이 각은 영상의 라인

방향에 대하여 정의된다. 또한,

와

는 보조자료에서 제공하는 제 2 관측각으로, 선형 스캐닝 영상의 경우 영상의 픽셀

방향으로 정의되고,

와

는 상기 제1 관측각과 제 2 관측각이 다르기 때문에 발생하는 오차 관측각으로 영상의 라인과 픽셀에 대한 함수로 정의된다. here,

Is

and

It is a variable used to convert the unit of to distance from the ground, and the image point

Is defined as the image coordinate, and the ground point for this image point.

Is defined as the Earth-Centered Earth-Fixed Coordinate System (ECEF).

Wow

Ground point

And location

The first observation angle, defined by the vector in between, which is the line of the image

It is defined with respect to the direction. Also,

Wow

Is the second observation angle provided by the supplementary data, and in the case of a linear scanning image, the pixel of the image.

Direction,

Wow

Is an error observation angle generated because the first observation angle and the second observation angle are different, and are defined as a function of lines and pixels of the image.

와

는 지상점, 촬영위치 및 촬영자세에 의하여 하기와 같이 정의된다.

Wow

Is defined as follows by the ground point, the photographing position and the photographing posture.

는 지구중심에서 촬영위치까지의 거리를 나타내며,

,

,

는 회전행렬

의 요소이고,

,

와

는 회전행렬

의 요소이다. here,

Indicates the distance from the center of the earth to the shooting location,

,

,

Rotation matrix

Is an element of

,

Wow

Rotation matrix

It is an element of.

The two rotation matrices are defined by Equations 5 and 6 below.

는 행렬

의 전치행렬이고,

는 국지궤도 표계(Local Orbital Coordinate System)를 자세좌표계(Attitude Coordinate System)로 변환하는 회전행렬이며,

는 국지궤도 좌표계를 지심좌표계로 변환하는 회전행렬이다. 보다 자세한 사항은 출원번호 제10-2006-26655호 "LOS벡터 조정 모델을 이용한 영상의 기하보정 방법 및 그 장치"를 참조한다.here,

Is a matrix

Transpose of,

Is a rotation matrix that converts the Local Orbital Coordinate System into an Attitude Coordinate System,

Is a rotation matrix that transforms a local orbit coordinate system into a center coordinate system. For more details, refer to Patent Application No. 10-2006-26655, "Geometric Correction Method and Apparatus for Image Using LOS Vector Adjustment Model".

와

를 0으로 가정하고, 대응점에 대하여 입체영상으로부터 제 1지상좌표

를 추출한다.Error Observation Angle in Equations 1 and 2

Wow

Is assumed to be 0, and the first ground coordinate from the stereoscopic image for the corresponding point.

Extract

개 이상의 영상에서 지상점의 결정은 지상점의 초기치 (

,

,

)로부터 하기의 수학식에 의하여 계산할 수 있다. 상기 지상점의 초기치는 보조자료로부터 추출할 수 있다. 이러한 과정은 수치사진측량분야에서 일반적으로 사용되고 있는 방법이다.

In more than one image, the ground point is determined by the initial value of ground point (

,

,

Can be calculated by the following equation. The initial value of the ground point may be extracted from auxiliary data. This process is a commonly used method in the field of digital photogrammetry.

는 잔차벡터이고,

는

크기의

,

와

에 대한 편미분 계수 행렬로 다음과 같이 정의한다.here

Is the residual vector,

Is

Size

,

Wow

The partial derivative coefficient matrix for is defined as

은

크기를 지니는

,

와

의 근사값에 대한 보정량 벡터로 하기 수학식 9에 의해서 정의된다.vector

silver

Sized

,

Wow

A correction amount vector for an approximation of is defined by the following equation (9).

는 관측치과 계산치간의 차이량 벡터로 하기 수학식 10과 같다.vector

Is the difference vector between the observed value and the calculated value.

,

와

의 근사치에 대한 보정량은 하기의 행렬식으로 구한다.From the matrix constructed as above

,

Wow

The correction amount for the approximation of is obtained by the following determinant.

The obtained correction amount is added to the initial value, and then substituted into Equation 7 using the summed value to obtain the correction amount. This process continues until the following equation (12) is obtained.

은 일반적으로

정도를 이용한다.Experiments have shown that this process usually requires fewer than four repetitions,

Is generally

Use degree.

는 불완전한 보조정보에 의하여 오차를 지니게 된다. 비록 제 1 지상좌표

와 원 지상좌표

의 차이가 크더라도 이 오차는 대응점에서 일정한 차이를 보인다. First ground coordinate determined through the above process

Has an error due to incomplete auxiliary information. 1st ground coordinate

And circle ground coordinates

Although the difference is large, this error shows a constant difference at the corresponding point.

FIG. 2 is a graph illustrating an embodiment of an error of first ground coordinates calculated from a SPOT 3 image, which is a linear scanning image, for 35 corresponding points. The results showed that the first ground coordinates had very large errors of -532m, -473m, and -153m in the X, Y, and Z directions, respectively, but the variance was almost constant at 4.41m, 7.14m, and 8.99m, respectively. Through this, it can be seen that the first ground coordinates have a certain difference from the original coordinates.

를 추출하는 단계이다. 본 단계는 제 1 지상좌표가 원 지상좌표와 차이가 크더라도 그 차이는 일정하다는 것에 근거하여 진행된다.The second ground coordinate extraction step 40 is to obtain a correlation coefficient map between the first ground coordinates determined above and the previously produced numerical elevation data to obtain a second ground coordinate.

Extracting step. This step is carried out on the basis that the difference is constant even if the first ground coordinate differs from the original ground coordinate.

Hereinafter, a method of extracting a second ground coordinate will be described with reference to the accompanying drawings.

3 is a flowchart illustrating a procedure flow of a method of extracting a second terrestrial coordinate according to the present invention.

The first coordinate transformation 41 is a step of converting the extracted first ground coordinates into a coordinate system possessed by the numerical elevation material 50. In one embodiment, SRTM DEM, a numerical elevation data provided worldwide, is provided in a geodetic coordinate system. When using the SRTM DEM, the first ground coordinates given in the center coordinate system are converted into longitude and latitude coordinates.

The correlation coefficient map preparation 42 is a step of preparing a correlation coefficient map by calculating a correlation coefficient from an altitude value of the first ground coordinate converted into a coordinate system of the numerical elevation data and an elevation value of the numerical elevation data.

라 한다. 상술한 바와 같이 제 1 지상좌표는 원 지상좌표와 일정한 차이를 지니고 있다. 그러므로 변환된 제 1 지상좌표는 수치표고자료와

(경도)방향,

(위도)방향 및

(고도)방향으로 일정한 오차

를 지닌다. 변환된 제 1 지상좌표의 수평좌표

에 해당하는 고도를 수치표고자료로부터 추출한다. 만약 제 1 지상좌표의 수평좌표가 원 지상좌표의 수평좌표와 다르다면 각 대응점에서 상기 수치표고자료로부터 추출된 고도와 변환된 제 1 지상좌표의 고도(

)의 차이는 일정하지 못하게 된다. 이는 원 지상좌표의 수평좌표가 아닌 곳에서 고도를 추출하였기 때문에 변환된 제 1 지상좌표의 고도(

)와는 다른 값을 지니게 되고, 각 대응점에 대해서 일정한 오차를 지니지 못하기 때문이다. 그러나 변환된 제 1 지상좌표의 수평좌표

에 일정한 오차가 더해진 좌표

가 원 지상좌표와 동일하다면 이때에 수치표고자료에서 고도를 추출하였다면, 이 고도는 분명 변환된 제 1 지상좌표의 고도(

)와 일정한 오차를 지닐 것이다. 여기서, 일정한 오차를 지닌다는 것은 변환된 제 1 지상좌표의 고도(

)와 수치표고자료에서 추출된 고도가 모든 대응점에 대해 서로 동일한 차이를 보인다는 것이고, 이는 바로 상관계수가 높다는 것을 의미한다. The first ground coordinate converted by the first coordinate transformation 41 is

It is called. As described above, the first ground coordinates have a certain difference from the original ground coordinates. Therefore, the converted first ground coordinates

(Hardness) direction,

(Latitude) direction and

Constant error in the (altitude) direction

Has Horizontal coordinate of converted first ground coordinate

The altitude corresponding to is extracted from the numerical elevation data. If the horizontal coordinate of the first ground coordinate is different from the horizontal coordinate of the original ground coordinate, the altitude extracted from the numerical elevation data at each corresponding point and the altitude of the converted first ground coordinate (

) Is not constant. Since the altitude is extracted from the non-horizontal coordinate of the original terrestrial coordinate,

This is different from), and it does not have a constant error for each corresponding point. However, the horizontal coordinate of the converted first ground coordinate

Coordinates with constant error

If the altitude is the same as the original ground coordinate, then if the altitude is extracted from the numerical elevation data, then this altitude is clearly the altitude of the transformed first ground coordinate (

) And will have a certain error. Here, having a constant error means that the altitude of the transformed first ground coordinate (

) And the altitude extracted from the numerical elevation data show the same difference for all the corresponding points, which means that the correlation coefficient is high.

)와 Y방향의 최소오차(

)를 -1500m로, 그리고 X방향의 최대오차(

)와 Y방향의 최대오차(

)를 1500m로 결정한다. 또한 SPOT 3 위성의 공간해상도는 10m이므로 오차의 간격은 10m로 결정한다.First, the minimum and maximum values of the error are set according to the accuracy of the extracted auxiliary data, and the error interval is set. As an embodiment of the present invention, since the SPOT 3 satellite generally has an error of 500 m to 1000 m, the minimum error in the X direction in consideration of the worst case (

) And the minimum error in the Y direction (

) To -1500m and the maximum error in the X direction (

) And the maximum error in the Y direction (

) Is 1500m. In addition, the spatial resolution of the SPOT 3 satellite is 10m, so the error interval is determined to be 10m.

를 계산한 후, 상기 수평좌표에 해당하는 고도를 수치표고자료에서 추출하고, 상기 수치표고자료에서 추출된 고도와 제 1 지상좌표의 고도(

)의 상관계수를 형성한다. 이 상관계수는 X와 Y방향으로의 최소오차에서 최대오차까지 오차간격으로 계산되어 2차원의 상관계수맵을 계산한다. A correlation coefficient map is produced at an error interval from the determined minimum error to the maximum error. Horizontal coordinates obtained by adding an error to the horizontal coordinates of the first ground coordinates converted for all the corresponding points extracted in the corresponding point extracting step 20.

After calculating the, the altitude corresponding to the horizontal coordinates is extracted from the numerical elevation data, the altitude and the altitude of the first ground coordinates extracted from the numerical elevation data (

Form a correlation coefficient of The correlation coefficient is calculated as the error interval from the minimum error to the maximum error in the X and Y directions to calculate a two-dimensional correlation coefficient map.

4 is an embodiment of a correlation coefficient map calculated using a SPOT 3 satellite, which is a linear scanning satellite image. As shown in the figure, the correlation coefficient map had a higher correlation coefficient closer to the original ground coordinate, and the maximum correlation coefficient was 0.991.

)와 Y방향의 오차(

)를 결정한다. 상기와 같이 결정된 수평좌표

로부터 수치표고자료에서 추출한 고도와 제 1 지상좌표의 고도(

)를 이용하여 고도오차(

)를 결정한다.The maximum correlation coefficient positioning 43 determines a maximum value in the produced correlation coefficient map, and further interpolates in sub-pixel units in the X and Y directions from the position where the correlation coefficient is maximum. Accurately determine the position where the correlation coefficient is maximum, and determine the error in the X direction (

) And the error in the Y direction (

Is determined. Horizontal coordinates determined as above

Altitude extracted from digital elevation data and altitude of first ground coordinate (

) Using the altitude error (

Is determined.

FIG. 5 is an embodiment in which a position having a maximum value of a correlation coefficient map is interpolated by using a SPOT triad that is a linear scanning satellite image. The interpolation was performed at 0.05 pixels using spline interpolation.

)는 -847.5m이었으며, Y방향의 오차(

)는 92.5m이었고, 고도오차(

)는 163.62m이었다.The error in the X direction determined from this (

) Was -847.5 m, and the error in the Y direction (

) Was 92.5m and the altitude error (

) Was 163.62 m.

), Y방향의 오차(

) 및 고도오차(

)를 X와 Y방향에 따라 오차량을 다르게 보정하는 것이다. 상기 최대상관계수 위치 결정 단계(43)에서 추출한 오차는 X와 Y 방향으로의 평균오차이다. 하지만, 이들 오차는 수평방향(X, Y)에 대해 그 오차량이 다르다. The error amount correction 44 is an error in the X direction extracted in the maximum correlation coefficient positioning step 43

), The error in the Y direction (

) And altitude error (

) Is to compensate the error amount differently according to X and Y direction. The error extracted in the maximum correlation coefficient positioning step 43 is an average error in the X and Y directions. However, these errors have different amounts of errors with respect to the horizontal directions (X, Y).

) 분포의 일례를 보인다. 상기 도면에서 볼 수 있듯이 Y방향으로의 오차(

)는 X와 Y에 대하여 오차량이 변한다는 것을 알 수 있다. 상기와 같은 이유는 영상에서 각 픽셀들이 다른 관측각을 지니기 때문에 발생하는 것으로 이를 보정하지 않으면, 절대적인 오차는 크지 않을지 모르나, X와 Y 방향을 따라 오차가 전파되는 양상을 지니게 된다. 6 shows the Y-direction error of X and Y for the SPOT 3 satellite, which is a linear scanning satellite image.

) Shows an example of distribution. As can be seen in the drawing, the error in the Y direction (

It can be seen that the error amount changes for X and Y. The reason for this is that each pixel in the image has a different viewing angle. If not corrected, the absolute error may not be large, but the error propagates along the X and Y directions.

,

및

를 계산한다. 각 구획별로 결정된 오차량을 통하여 X와 Y방향에 따라 다른 오차량을 적용한다. 이 과정은 상기 상관계수맵 제작 단계(42)와 최대상관계수 위치결정 단계(43)에서 추출한 오차의 정확도에 따라 오차의 최소값과 최대값을 설정하고, 오차의 간격을 설정한다. In order to correct this, unlike the correlation coefficient map using all the corresponding points in the correlation coefficient map production step 42, the corresponding points are divided into several sections for the X and Y directions as shown in FIG.

,

And

Calculate Different error amounts are applied in the X and Y directions through the error amounts determined for each section. In this process, the minimum and maximum values of the error are set according to the accuracy of the error extracted in the correlation coefficient map preparation step 42 and the maximum correlation coefficient positioning step 43, and the error interval is set.

)와 Y방향의 최소오차(

)를 -20m로, 그리고 X방향의 최대오차(

)와 Y방향의 최대오차(

)를 20m로 결정할 수 있고, 원하는 정확도에 따라 오차의 간격을 결정할 수 있다. 본 실시예 에서는 2m로 결정하였다.As an example, the accuracy of the error extracted in the correlation coefficient map production step 42 and the maximum correlation coefficient positioning step 43 for the SPOT 3 satellite is within 30 m.

) And the minimum error in the Y direction (

) To -20m and the maximum error in the X direction (

) And the maximum error in the Y direction (

) Can be determined as 20m, and the interval of error can be determined according to the desired accuracy. In this example, it was determined to be 2m.

), Y방향의 오차(

) 및 고도오차(

)를 변환된 제 1 지상좌표

에 추가하여 오차를 보정한 지상좌표

를 지심좌표계로 변환하는 단계이다. 지심좌표로 변환된 좌표는 제 2 지상좌표

가 된다. 선형 스캐닝 위성영상인 SPOT 3위성을 이용한 실시예에서 제 2 지상좌표는 제 1지상좌표와 X, Y와 Z 방향에 대하여 각각 평균 -527m, -470m 및 -159m의 오차가 보정되었다. 이는 제 1지상좌표와 원 지상좌표의 평균오차와 거의 동일한 값이다. The second coordinate transformation 45 is an error in the X direction in which the error in the X and Y directions is corrected in the error amount correcting step 44.

), The error in the Y direction (

) And altitude error (

Transformed first ground coordinate

Ground coordinates corrected for errors in addition to

This step converts the centroid coordinate system to. Coordinates converted to geocentric coordinates are second ground coordinates.

Becomes In the embodiment using the SPOT terrestrial satellite, which is a linear scanning satellite image, the second ground coordinates have an error of -527m, -470m, and -159m on the first ground coordinate and the X, Y, and Z directions, respectively. This is almost equal to the mean error of the first ground coordinate and the original ground coordinate.

FIG. 8 is a diagram illustrating the difference between the second ground coordinate and the original ground coordinate determined using the SPOT 3 satellite, which is a linear scanning satellite image, and it can be seen that the second ground coordinate has an error within 20 m. In particular, the second ground coordinate error (RMSE) was less than 8m in all directions. This is a very accurate result considering that the spatial resolution of the used image is 10m and the spatial resolution of the used digital elevation data is 90m. The precision of the second ground coordinates depends on the spatial resolution and accuracy of the numerical elevation data used and on the spatial resolution of the image used.

Hereinafter, a method of correcting a photographing position and posture will be described with reference to the accompanying drawings.

및 자세

를 개선하는 단계이다. The photographing position and posture improvement step 60 is a photographing position of the auxiliary data of the image and ancillary data extraction step 10 using the second ground coordinates calculated in the second ground coordinate extraction step 40.

And posture

Step to improve it.

The present invention proposes a method which can be applied regardless of the accuracy of the photographing position by first correcting the photographing position and posture before correcting the error observation angle.

)은 촬영위치오차에 의한 위치오차관측각(

)와 자세각의 오차에 의한 자세오차관측각(

)에 의해서 하기의 수학식 13과 14와 같이 표현할 수 있다.Error observation angles defined in Equations 1 and 2 from FIG.

) Indicates the position error observation angle due to the shooting position error (

) And posture error observation angle due to error of posture (

) Can be expressed as in Equations 13 and 14 below.

와

는 각각

의 X와 Y방향성분이고,

와

는 각각

의 X와 Y방향성분이다. here

Wow

Are each

In the X and Y directions,

Wow

Are each

Are the X and Y direction components.

)은 사인의 법칙에 따라 하기의 수학식 15로 표현할 수 있다.Position error observation angle (

) Can be expressed by Equation 15 below according to the law of sine.

는 원래의 촬영위치와 보조자료에 제공된 촬영위치사이의 거리이며,

는

와

의 수평성분사이의 각이고,

은 보조자료에 제공된 촬영위치와 지상위치사이의 거리이다.here,

Is the distance between the original shooting position and the shooting position provided in the auxiliary material,

Is

Wow

Is the angle between the horizontal components of

Is the distance between the shooting position and the ground position provided in the ancillary data.

는 0에 가깝기 때문에 수학식 15는 하기 수학식 16과 같이 근사화 될 수 있다.

Since E is close to 0, Equation 15 may be approximated as Equation 16 below.

의 X와 Y 성분(

,

)은 하기의 수학식 17과 수학식 18과 같이 정의된다.therefore

X and Y components of

,

) Is defined as in Equations 17 and 18 below.

와

는

의 X와 Y성분이고,

와

는

의 X와 Y성분이며,

와

는

의 X와 Y성분이다. here,

Wow

Is

Are the X and Y components of

Wow

Is

Are the X and Y components of

Wow

Is

X and Y components of.

를 이용하면, 하기의 수학식 19와 수학식 20과 같이 정의된다.Location Error from Drawing 10

By using Equation 19, Equation 19 and Equation 20 are defined.

)의 X와 Y 성분(

,

)은 자세오차

를 이용하여 하기 수학식 21과 수학식 22와 같이 정의된다.Also, posture error measurement angle (

X and Y components of)

,

) Is a posture error

By using Equation 21 and Equation 22 are defined.

와

는 피치각 오차와 롤각 오차이고,

와

는 요각 오차

의 X와 Y성분이다.

와

는 도면 11(ㄷ)으로부터 사인의 법칙을 이용하여 정의할 수 있다.

를 정의하기 위하여 사인의 법칙을 적용하면 하기의 수학식 23과 같이 정의된다.here

Wow

Is the pitch angle error and roll angle error,

Wow

Is the yaw error

X and Y components of.

Wow

Can be defined using the law of sine from FIG.

Applying the law of sine to define the equation is defined as in Equation 23 below.

이고,

는 위성의 고도이며,

는 지표면의 고도이다.here

ego,

Is the altitude of the satellite,

Is the elevation of the earth's surface.

는 0에 가깝고,

와

는

와 근사적으로 같다. 그러므로 수학식 23은 하기 수학식 24와 같이 근사화될 수 있다.

Is close to 0,

Wow

Is

Is approximately equal to Therefore, Equation 23 may be approximated as Equation 24 below.

와

은 도면 11(ㄴ)으로부터 하기와 같이 정의된다.here

Wow

Is defined as follows from Fig. 11 (b).

는 하기 수학식 27과 같이 근사화된다.finally

Is approximated as in Equation 27 below.

와 동일한 절차를 거쳐 하기 수학식 28과 같이 정의된다. Also

It is defined as in Equation 28 through the same procedure as

)의 X와 Y 성분(

,

)은 자세오차

를 이용하여 하기 수학식 29와 수학식 30과 같이 정의된다.Finally, posture error observation angle (

X and Y components of)

,

) Is a posture error

By using the equations (29) and (30) are defined as follows.

와 자세오차

부터 결 정되는 오차관측각(

)은 하기 수학식 31과 수학식 32로 정의된다.Finally, shooting position error

And posture error

Error observation angle determined from

) Is defined by the following equation (31) and (32).

를 상기 수학식 1과 수학식 2에 대입하여 각 제 2 지상좌표에 대한 X와 Y방향의 오차관측각

와

를 계산한다. 상기 계산된 오차관측각을 수학식 31과 수학식 32에 대입하여 촬영위치오차

와 자세오차

를 계산한다. 여기서 주의해야 할 점은 일반적으로 선형스캐닝 영상에서

와

를 구분할 수 없고,

와

를 구분할 수 없다는 것이다. 이와 같은 이유는 선형스캐닝 영상이 주로 비행방향의 수직으로 CCD를 배열하기 때문에 발생한다. 즉, 수학식 31과 수학식 32에서

값의 변화가 매우 작아 상수로 인식할 수 있기 때문이다. 그러나 일반적인 선형스캐닝 영상에서

와

에 비해

와

의 차이는 미미하기 때문에

와

으로 가정할 수 있 다. 결국 수학식 31와 수학식 32를 이용하여

로 가정하고, 촬영위치오차

와 자세오차 중

를 개선한다. 만약, 선형스캐닝영상에서

와

의 변화가 상수로 인식할 수 없을 만큼 크다면, 모든 자세오차를 개선할 수 있다. Second ground coordinates extracted in the second ground coordinates extraction step 40

Is substituted into Equation 1 and Equation 2, and error observation angles in the X and Y directions with respect to the second ground coordinates, respectively.

Wow

Calculate Imaging position error by substituting the calculated error observation angle into Equation 31 and Equation 32

And posture error

Calculate One thing to note here is that for linear scanning images

Wow

Cannot be distinguished,

Wow

Is indistinguishable. The reason for this is that the linear scanning image mainly arranges the CCD in the vertical direction of the flight direction. That is, in Equation 31 and Equation 32

This is because the change in value is so small that it can be recognized as a constant. But in a typical linear scanning image

Wow

Compared to

Wow

Because the difference is small

Wow

Can be assumed. Finally, using Equation 31 and Equation 32

Assume that the shooting position error

And posture error

To improve. If, in the linear scanning image

Wow

If the change in is large enough to be unrecognizable as a constant, all posture errors can be improved.

If the shooting position error and posture error of the auxiliary material is improved by the above process, the original shooting position and the shooting posture are closer to each other, and the geometric correction is performed when the accuracy of the shooting position of the auxiliary material generated by the prior art is very low. Overcomes the disadvantage of very low precision.

The external index performing step 70 uses the second ground coordinates of the second ground coordinate extraction step 40 and the improved shooting position and posture of the shooting position and posture improvement step 60 to calculate the error observation angle. By calibrating, external coordinates are given to assign the ground coordinates to each image coordinate of the image.

)은 한국특허 출원번호 제10-2006-26655호 "LOS벡터 조정 모델을 이용한 영상의 기하보정 방법 및 그 장치"의 LOS 벡터 조정 모델에서 제안한 방법과 같이 하기의 수학식 33과 34에 의해서 계산할 수 있다.Error observation angles of Equations 1 and 2

) Can be calculated by the following equations 33 and 34, as proposed in the LOS vector adjustment model of Korean Patent Application No. 10-2006-26655 "Method of Correcting Image Using LOS Vector Adjustment Model and Its Apparatus". have.

,

,

,

,

및

은 지상기준점에 의해 보정되는 외부표정모델 계수이다. 일반적으로

과

을 사용하는 것이 바람직하다. 즉, 선형스캐닝 영상의 라인방향(

)으로는 1차방정식을 픽셀방향(

)으로는 2차방정식을 사용하는 것이 바람직하다. 픽셀방향(

)에 2차방정식을 사용하는 것이 바람직한 이유는 영상의 픽셀방향에 따라 변하는 관측각

와

의 변화가 2차 방정식에 근사하기 때문이다. here

,

,

,

,

And

Is the external expression model coefficient corrected by the ground reference point. Generally

and

Preference is given to using. That is, the line direction of the linear scanning image (

) Gives the first equation in the pixel direction (

It is preferable to use a quadratic equation as). Pixel direction (

It is preferable to use quadratic equations for the observation angle which varies with the pixel direction of the image.

Wow

This is because the change in is close to the quadratic equation.

After this step, the image has information about ground coordinates. Thus, it is possible to produce a 3D image based on this, to produce an orthoimage by reconstructing pixel positions, and to produce a digital topographical map, digital elevation data, and the like.

In the above description of the present invention with reference to the accompanying drawings has been described a geometric correction method of the image by adjusting the LOS vector having a specific step and components and the device, but the present invention can be variously modified and changed by those skilled in the art And, such variations and modifications should be construed as falling within the protection scope of the present invention.

According to the automatic image geometry correction method of the linear scanning image of the present invention as described above,

First, the ground reference point can be automatically extracted from the digital elevation model (DEM) to assign accurate ground coordinates to each image coordinate of the image.

Second, the cost and time required for image geometry correction can be significantly reduced.

Third, it is possible to improve the accuracy of the extracted ground coordinates,

Fourth, the orthogonal image, image map, digital topographic map, and work process of digital elevation data can be simplified and shorten working time.

Claims (13)

(a) extracting two or more different images of the same ground surface and auxiliary data for the images; (b) extracting corresponding points through image matching from two or more different images; (c) a first ground coordinate from the correspondence point and the ancillary data;

Extracting; (d) the second ground coordinate using the correlation coefficient between the first ground coordinate and the numerical elevation data;

Extracting; (e) performing an external calibration to correct distortion of the image by assigning the ground coordinates to each image coordinate of the image by using the second ground coordinates. Way. The method according to claim 1, After step (d) and before step (e), Shooting position of the auxiliary data using the second ground coordinate

And posture

Automatic image geometry correction method of the linear scanning image, characterized in that it further comprises the step of improving. The method according to claim 1, In step (d), A first coordinate conversion step of converting the first ground coordinates into a coordinate system having the numerical elevation data; Producing a correlation coefficient map using the converted first ground coordinates and an elevation value of the numerical elevation data; Determining a maximum correlation coefficient position in the produced correlation coefficient map; Correcting the error amount extracted in the determining of the maximum correlation coefficient position; And a second coordinate conversion step of inversely converting the coordinates of which the error amount is corrected into a coordinate system before conversion. The method according to claim 1 or 2, The auxiliary data of step (a) may include information about the position of the photographing apparatus, the speed information, the attitude information of the photographing apparatus, the photographing time of the center line of the image, the photographing time per line of the image, the photographing angle, etc. Automatic image geometry correction method of linear scanning image. The method according to claim 1 or 3, The step (b) includes calculating a correlation coefficient with respect to different images to find a place where the correlation coefficient is the highest. The method according to claim 2, Shooting position of the auxiliary data using the second ground coordinate

And posture

Steps to improve it,

and

Error observation angle using

,

Calculate the above

Is

and

It is a variable used to convert unit of to distance from the ground, and the image point

Is defined as the image coordinate, and the ground point for this image point.

Is defined as the Earth-Centered Earth-Fixed Coordinate System (ECEF),

Wow

Ground point

And location

The first observation angle, defined by the vector in between, which is the line of the image

Defined with respect to the direction,

Wow

Is the second observation angle provided by the supplementary data, and in the case of linear scanning images, the pixel of the image.

Is defined as a direction,

Wow

Is an error observation angle generated because the first observation angle and the second observation angle are different, and are defined as a function of lines and pixels of the image; The error observation angle

and

Position error

And posture error

Calculate the above

Wow

Is the yaw error

X and Y components of,

Is the angle from the vertical component of the shooting position and the ground position (

), Rx and Ry are automatic image geometry correction methods for a linear scanning image, wherein R is the X and Y components of the distance between the photographing position and the ground position. The method according to claim 2 or 3, In step (e), And correcting the error observation angle by using the second ground coordinates, the improved shooting position, and the shooting position, thereby assigning the ground coordinates to each image coordinate of the image. The method according to claim 3, Producing a correlation coefficient map using the converted first ground coordinates and the altitude value of the numerical elevation data, According to the accuracy of the auxiliary data extracted in step (a), the minimum and maximum values of the error are set, and the interval of the error is set, and then the correlation coefficient map is produced from the error interval from the minimum value of the error to the maximum value of the error. Automatic image geometry correction method for a linear scanning image, characterized in that. The method according to claim 3, The determining of the maximum correlation coefficient position in the produced correlation coefficient map, Determining a maximum value in the produced correlation coefficient map; Performing interpolation in sub-pixel units in the X and Y directions from a position where the correlation coefficient is maximum; Determining a position where a correlation coefficient is maximum by performing the interpolation; Using this crystal, the error in the X direction (

) And the error in the Y direction (

Determining; And determining the altitude error using the altitude extracted from the digital elevation data and the altitude of the first ground coordinate using the error. The method according to claim 3, Correcting the amount of error extracted in the step of determining the maximum correlation coefficient position, An automatic image geometry correction method of a linear scanning image, characterized in that the error amount in the X direction, the error in the Y direction, and the altitude error are differently corrected according to the X and Y directions. The method according to claim 1 or 3, The numerical elevation data may be provided by any one of the elevation data extracted using a shuttle radar topography mission (SRTM) digital elevation model (DEM), a GTOPO-30, or a TIN grid and a lidar data. Automatic geometric correction method. The method according to claim 3, The first coordinate transformation step of converting the first ground coordinates into a coordinate system having the numerical elevation data, When the digital elevation data provided by STRM DEM is used as the digital elevation data, the automatic image geometry correction method of the linear scanning image, characterized in that the given coordinates are converted into longitude coordinates. delete

Images