KR100520275B1 - Method for correcting geometry of pushbroom image using solidbody rotation model - Google Patents

Abstract

 The present invention relates to a geometric correction technology of a linear scanning image, and more particularly, to linearly scan a linear scanning image such as an aerial photograph or a satellite image photographed by a photographing apparatus such as an airplane or a satellite, from the image auxiliary data. , Y, Z), speed information of the shooting device (Vx, Vy, Vz), center line shooting time of the image, shooting time per line of the image, posture information (yaw, pitch and roll) of the shooting device, and X, Y direction The sensor model was established by using the photographing angle, etc., the rotation of the photographing device in the Y-axis direction from the earth center based on the local orbit coordinate system, The present invention relates to a technique for correcting geometric distortion of an image using a photographing device rotation model, which is a corrected sensor model at a position.

In order to achieve the above object, a geometrical correction method of a linear scanning image using the photographing apparatus rotating model includes: a first step of acquiring image auxiliary data of a linear scanning image; Selecting two or more accurate ground control points obtained by GPS surveying, and using the following equation (13) and equation (17), the parameters of the photographing instrument rotation model expressing the relationship between the image coordinates and the ground coordinates And Determining a second step; Determined by the image auxiliary data of the first step and the second step And And a third step of performing external expression of the image by applying the following equations (10) and (11), which are the photographing mechanism rotation models.

According to the present invention, two or more accurate ground reference points obtained by GPS surveying can effectively correct distortion of an image from a photographic instrument rotation model, and use it to accurately correct digital topographic maps, image maps, and digital elevation data. Maps can be made.

Description

Geometric correction method of linear scanning image using photographing device rotation model when producing digital map by GPS surveying {METHOD FOR CORRECTING GEOMETRY OF PUSHBROOM IMAGE USING SOLIDBODY ROTATION MODEL}

The present invention relates to a geometric correction technology of a linear scanning image, and more particularly, to linearly scan a linear scanning image such as an aerial photograph or a satellite image photographed by a photographing apparatus such as an airplane or a satellite, from the image auxiliary data. , Y, Z), speed information of the shooting device (Vx, Vy, Vz), center line shooting time of the image, shooting time per line of the image, posture information (yaw, pitch and roll) of the shooting device, and X, Y direction The sensor model was established by using the photographing angle, etc., the rotation of the photographing device in the Y-axis direction from the earth center based on the local orbit coordinate system, The present invention relates to a technique for correcting geometric distortion of an image using a photographing device rotation model, which is a corrected sensor model at a position.

In addition to aerial photographs taken over the past several decades, satellite images produced to observe the earth's surface can be divided into satellites with passive sensors and satellites with active sensors. The most common satellites with passive sensors are SPOT, IRS, ALOS and QUICKBIRD. The most common satellites with active sensor are RADARSAT, JERS, ERS and ENVISAT. These satellite images are used in numerous applications such as image mapping, digital elevation data production, and digital mapping, which are related to the observation of the surface of the earth. It is a very important issue.

There are three general methods to calibrate the geometry of a conventional linear scanning image. First, it is extended by the collinear condition equation which is widely used in aerial survey. This method assumes an image taken by a linear scanning image as a single picture having different positions and postures for each line, and approaches the polynomial where the positions and postures for each line change along the lines. The second method is a method that is useful when there is no position, speed, and posture information of a photographing apparatus. This method, also called DLT (Direct Linear Transform), is better applied to aerial photography. The third method, which is applied to satellites, is called the orbital parameter method, which uses the orbital element assuming that the satellites move in a Kepler orbit and follow a well-defined ellipsoid.

These methods have advantages and disadvantages, and are used in different ways depending on the intended purpose. If the satellite header information is not available, the DLT method will be used, and when performing large-scale data production, many ground control points are required by GPS surveying, but do not rely heavily on the characteristics of the ground surface and satellite header information. An extended bundle adjustment method that can extract location information is used. However, it is necessary to extract accurate location information using only the exact ground reference points obtained from only a few GPS surveys.In order to solve this problem, the geometric distortion of the image is minimized by using the information provided in the image assistance data. There is a need for correction.

The present invention is to solve the above problems, to establish a photographing mechanism rotation model as a sensor model showing the relationship between image coordinates and ground coordinates, which is a parameter of the imaging mechanism rotation model And It is an object of the present invention to determine the geometric coordinates of an image by correcting the geometric distortion of the image by determining the using two or more ground reference points and then performing external expression using the photographing mechanism rotation model.

      Geometric correction method of the linear scanning image using the photographing mechanism rotation model of the present invention for achieving the above object,

A first step of acquiring image auxiliary data of the linear scanning image;

Selecting two or more accurate ground control points by GPS surveying and using the following equations (13) and (17), And Determining a second step;

Determined by the image auxiliary data of the first step and the second step And And a third step of performing external expression of the image by applying the following equations (10) and (11), which are the photographing mechanism rotation models. Here, the external coordinate means the ground coordinate (i) in the image coordinate (i, j) of the image. , , To say).

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

1 is a schematic process flow diagram according to the present invention. Referring to the present invention, the present invention acquires auxiliary data of an image (S10), selects an accurate ground reference point by GPS surveying (S20), and then determines parameters using it (S20). S30), by extracting the image coordinates (S40) by applying the auxiliary data and the parameters and the image coordinates to the photographing instrument rotation model (S50) is performed a procedure of performing an external coordinate to give the ground coordinates to the image coordinates (S60).

FIG. 2 represents the relationship between the photographing apparatus, the earth, and the photographing position under the assumption that the photographing apparatus moves along a well-defined ellipsoid when viewed in an ECI coordinate system (Earth-Centered Inertia Coordinate System). Same as 1. For reference, the photographing apparatus generally refers to a satellite, but includes an aircraft that has a similar motion to the satellite.

Equation 1

here, Location of the shooting mechanism, Is the shooting position on the earth, Expresses the shooting direction (LOS: Line-Of-Sight) in the photographing apparatus, Represents a parameter.

But what we usually know And the value of The value of is expressed in Earth-Centered Earth-Fixed Coordinate System, Is represented by the Attitude Coordinate System. In order to represent each variable of Equation 1 in the same coordinate system, we use a coordinate transformation matrix that converts the photographing instrument attitude coordinate system into a local orbital coordinate system. ) And coordinate transformation matrix for converting the geocentric coordinate system to the local orbit coordinate system ( ) Must be defined. The definitions are as shown in Equations 2 and 3 below.

Equation 2

Equation 3

here, Is the roll angle, Is the pitch angle, Is the plinth. Also, Represents the position vector of the shooting device, Represents the velocity vector of the photographing apparatus.

First, Equation 1 is re-expressed on the local orbit coordinate system using the transformation matrix represented by Equations 2 and 3, as shown in Equation 4.

Equation 4

here, , … Is Is an element of a matrix … Is Is an element of a matrix , , And , , Are each Wow It is an element of.

Through the equation 4, the image coordinates (i, j) and the ground coordinate ( , , The sensor model indicating the correspondence relationship is established. Where 'i' in the image coordinates is the number of lines in the horizontal direction and 'j' is the number of columns in the vertical direction.

Equation 5

Equation 6

Equation 7

The relationship between the image coordinate and the ground coordinate determined through the sensor models of Equations 5 and 6 is far lower than the desired accuracy. This is an auxiliary data of the image, which comes from inaccuracy (ie, error) such as the position, attitude, and speed of the satellite. The sensor model does not satisfy the error range required for image mapping and numerical elevation data. Therefore, in order to improve the accuracy of the relationship between the image coordinates and the ground coordinates, it is necessary to modify the trajectory of the imaging device to establish a new sensor model.

To this end, we established a rotating model of the photographing apparatus. Here, the rotation model of the photographing apparatus regards the photographing apparatus as a solidbody, and the position of the initial photographing apparatus based on the auxiliary data is set in the direction of the satellite in the direction of the satellite. Rotate it by as much as The sensor model at the modified trajectory (position) changed by the ratio of.

3 is a conceptual diagram illustrating a photographing mechanism rotation model. Referring to this,

Initial trajectory vector ( ) Size Increase by, vector Vector with On the plane When rotated as much as possible, that is, when the center of the earth is rotated on the local orbit coordinate system in the vertical direction (Y axis) in the direction of travel of the photographing apparatus, and the distance from the earth center to the photographing apparatus is changed (Z axis), The new position vector may be defined as in Equation 8 below.

Equation 8

here Is a function of time t and a number of lines i. This is because the shooting time per line of the image is constant, so the time t and the number of lines i are proportional. And May be expressed as Equation 9 below as a rotation problem about a line in a three-dimensional space.

Equation 9

here, Is a function of time t, , , Is Directional cosine component of the vector.

Wow In consideration of Equation 5 and Equation 6, the equation for the photographing apparatus rotation model is expressed as Equation 10 and Equation 11 below.

Equation 10

Equation 11

here, , … Is the transformation matrix of Equation 12 below It is an element of.

Equation 12

Equation 10 and Equation 11 are sensor model indicating a correspondence relationship between the image coordinate at the position of the newly-modified photographing apparatus and the ground coordinate of the photographed region.

The ground coordinates accurate to the image coordinates (i, j) of the image by the photographing apparatus rotation model ( , , In order to perform external expression And Should be determined.

parameter And The exact ground reference point is selected by GPS surveying, and is determined using the following equations (13) to (17) using the selected ground reference point. At this time, one ground reference point may be selected, but in this case, it is preferable to select two or more because the accuracy does not reach the level required for image mapping or digital elevation data. Also parameters And The range of error was found to be determined by calculating up to the first order polynomial for time t. This is due to the risk that the error increases due to overcorrection when calculating up to 4th and 5th order polynomials for time t.

Shooting mechanism rotation model parameter And The equation for obtaining is as shown in Equation 13 below.

Equation 13

here Is the residual, , And Is defined by Equations 14 to 16 below.

Equation 14

Equation 15

Equation 16

Where n is the number of ground control points and m is , The order used for n > m + 1.

From the following equation (17) Get the value, Equation 13 is repeated until And The value will be determined.

Equation 17

Parameters according to Equations 13 to 17 described above And The decision of is commonly used in numerical photogrammetry, such as the Taylor series expansion and the Newton-Rapson method. When the ground reference point is selected, the image coordinates (i, j) in the image corresponding to the ground reference point are determined. When GPS surveying is performed at the ground reference point, the ground coordinates of the point ( , , ) Is determined. Substituting these coordinate values into equation (10) and equation (11) And Is And It is a function of, which is a nonlinear equation, so we have a Taylor series expansion to linearize it. The result is Equation 13. However, immediately from Equation 13 And Can't be obtained by the Newton-Lobson method And Will be determined. In other words, And Is arbitrarily determined (usually set to 0) and substituted into Eq. And Is obtained. Substituting this value into Equation 13 again yields the new value. And Is obtained. Equation 13 is repeated as above And To determine the equation, Until is less than the reference value ( Becomes closer to 0).

Using the exact ground reference point obtained from the GPS survey as described above is a parameter of the imaging device rotation model from Equations 13 to 17 And Once the value is determined, the acquired video aid and the And By substituting the values into the above equations (10) and (11), which are photographing mechanism rotation models, the ground point ( , , However, since Equations 10 and 11 match three-dimensional positions with two-dimensional images, three-dimensional ground points ( , , ), The image point (i, j) of the two-dimensional image corresponding to the three-dimensional ground point can be calculated, but if the two-dimensional image point is given, the three-dimensional ground point ( , , ) Cannot be calculated, but we need one more image that contains the same region. In other words, if there is another video taken from another location in the same area, the same ground point ( , , Since each image has two image points within two images, and each image has Equations 10 and 11, a total of four equations are obtained. , , ) Can be given. To describe this process in detail, the ground reference point extracted by GPS is substituted into the first image, which is a parameter of the photographing device rotation model. Wow After determining the value of, apply the same to the second image, and extract the points of the same position in the first image and the second image and assign each image point into equation (10) and equation (11), We get a total of four equations, three unknown ground points ( , , ) Can be calculated from the above equation. Through this process, all ground coordinates (i, j) of the ground coordinates ( , , ), And moreover, it is possible to produce numerical elevation data indicating the height of the ground.

What has been described above has been described in detail the specific content of the present invention, the present invention is only one embodiment for explaining the geometric correction method of the linear scanning image using a photographing mechanism rotation model, the present invention is in the above embodiment Without departing from the gist of the invention as claimed in the following claims, those skilled in the art to which the invention belongs to the technical spirit of the present invention to the extent that various modifications can be made Will be said.

As described in detail above, according to the present invention, by applying the information provided from the accurate ground reference point and the image header obtained from only a few GPS surveys to the imaging device rotation model to extract the exact position information from the geometry of the linear scanning image It works. In addition, the use of less ground reference point than the conventional method, it is effective to shorten the work time by simplifying the work process, such as the image map production and digital elevation data production of the linear scanning image.

1 is a schematic process flow diagram in accordance with the present invention.

2 is a representation of the relationship between the photographing mechanism, the earth, and the photographing position in the ECI coordinate system.

3 is a conceptual diagram for explaining a photographing mechanism rotation model.

Claims (2)

A first step of acquiring auxiliary data of the linear scanning image; Selecting two or more accurate ground reference points by GPS surveying and using the following equations (13) and (17), And Determining a second step; Determined by the image auxiliary data of the first step and the second step And The third step of performing the external expression of the image by applying to the following Equation 10 and Equation 11, which is a photographing mechanism rotation model; linear scanning image using a photographing apparatus rotation model when producing a digital map by GPS survey Geometric correction method, [Where, the above And Is the ratio of the moving distance on the Z axis of the imaging device on the local orbit coordinate system ) And the angle of rotation on the Y axis ( ) to, , And Equation 13 is This, here Is the residual, , And Is defined by Equations 14 to 16 below, Equation 14 is ego Equation 15 is ego Equation 16 is This, Where n is the number of ground control points and m is , Order used for n≥m + 1, Equation 17 is ego, Equation 10 is ego Equation 11 is This, Where (i, j) is the image coordinate , , ) Is ground coordinates Is the vector of the shooting mechanism, , … Is an element of the transformation matrix generated by converting the geocentric coordinate system to the local orbit coordinate system. Represents the photographing direction in the photographing apparatus expressed in the local orbit coordinate system. The method of claim 1, Parameters determined in the second step And The geometric correction method of a linear scanning image using a photographic instrument rotation model when producing a digital map by GPS survey, characterized in that the calculated value up to the first polynomial for the time t.