Geometric correction method based on multispectral image

Abstract

The invention relates to a geometric correction method based on a multispectral image, and the method comprises the steps: carrying out the feature extraction of an original multispectral image, and generating ground control points; constructing a geometric correction model in which metadata is stored, and determining an affine transformation relationship between the ground control point and the metadata; and inputting the new multispectral image into the model for geometric correction, and outputting an orthoimage. Compared with the prior art, the method has the remarkable advantages that the geometric correction method is adopted, the orthorectification efficiency of the multispectral remote sensing image is remarkably improved, and a rigorous geometric data reference is provided for processing mass multispectral remote sensing data; moreover, block processing is added in the geometric correction process, the relative position and the absolute position between the images are recovered through connection points and control points which are overlapped and matched with the images, and the positioning and edge matching precision of the images is ensured.

Description

Technical Field

[0001] The invention relates to the technical field of image processing, and in particular to a method for geometric correction based on multispectral images. Background Art

[0002] Multispectral remote sensing earth observation is one of the important means for humans to obtain earth space information, and it plays an irreplaceable role in national economic construction and national defense construction. With the development of aerospace technology and sensor technology, a multi-level, multi-angle, all-round and all-weather global three-dimensional earth observation network is taking shape - combining high, medium and low orbits, coarse, medium and small coordination, and coarse, fine and precise resolution complementarity.

[0003] With the development of multispectral remote sensing earth observation technology, extracting information from multispectral remote sensing images, projecting multispectral remote sensing images into a fixed reference system and correcting the geometric distortion of the original images is usually called multispectral image geometric correction, so as to carry out geometric measurement, mutual comparison and composite analysis of image information; the errors generated at this stage will affect a series of subsequent analyses and decisions. Therefore, how to accurately project multispectral remote sensing images into a specified reference system and accurately eliminate the geometric deformation of the original images is a key technology in multispectral remote sensing image processing and application.

[0004] The problems in the study of geometric correction of multispectral remote sensing images are mainly concentrated on: the geometric distortion of remote sensing pixels is formed by the superposition of systematic errors (roll, pitch, yaw, etc.) and random errors (displacement of terrain fluctuations caused by deviation from sub-satellite observations), and the surface fluctuations covered by multispectral remote sensing images have the characteristics of randomness. Especially in the vast mountainous areas of my country, only the geometric errors of the image system caused by vibration, tumbling, etc. can be eliminated through polynomial transformation, while the random errors caused by terrain fluctuations need to be corrected pixel by pixel according to the information of the mountain altitude, height and side viewing angle. The inaccurate geometric positioning and the distortion errors of mountain terrain limit the application of multispectral images. The traditional geometric correction algorithm requires manual selection of control points, which is very time-consuming and has low accuracy, and it is difficult to meet the needs of massive remote sensing data processing. Therefore, the present invention designs a method for realizing automated geometric correction and orthorectification to obtain high-precision image positioning. Summary of the invention

[0005] The purpose of the present invention is to provide a method based on multispectral image geometric correction, which achieves the purpose of automatic geometric correction of multispectral images by capturing ground control points and using affine transformation of ground control points and metadata to realize radiation orientation.

[0006] The technical solution to achieve the purpose of the present invention is: A method for geometric correction of multispectral images, the method comprising: Extract features from original multispectral images and generate ground control points; Constructing a geometric correction model, in which metadata is stored, and determining the affine transformation relationship between ground control points and metadata; The new multispectral image is input into the model for geometric correction and the orthophoto is output.

[0007] Furthermore, the metadata includes data tags and attribute information of the original multispectral image, and the attribute information includes the acquisition date, band information, resolution, and geographic location of the original multispectral image.

[0008] Furthermore, the gradient value of each pixel in the original multispectral image is calculated, and the grayscale change measure of each window centered on each pixel after moving in the surrounding directions is calculated. Each grayscale change measure obtained is compared with the preset empirical threshold to determine the candidate points; the extreme value points are selected from the covariance matrix of the candidate points as ground control points.

[0009] Furthermore, the metadata is preprocessed, specifically including: extracting a geographic reference point in the metadata, creating a first window with a radius d centered on the geographic reference point; creating several search windows with a radius d centered on each ground control point in the original multispectral image; matching the first window with each search window, wherein the ground control point in the search window with the greatest matching similarity is the point with the same name as the geographic reference point.

[0010] Furthermore, the affine transformation relationship between the ground control points and the metadata is determined as follows: the regional matrix is ​​set, the length plane of the ground control points in the regional matrix is ​​obtained, and the affine transformation matrix is ​​generated by combining the length of the ground control points in the original multispectral image; the specific matrix expression is: Where: is the coordinate value of the origin of the original multispectral image in the new coordinate system, are the coordinates of the ground control points in the original multispectral image, are the coordinates of the ground control points in the regional matrix, , is the unit length of the two coordinate axes of the original coordinate system, expressed in the length unit of the corresponding axis of the regional matrix, The scale factor for the selected reference point of the area matrix, , The angles required to rotate the two coordinate axes of the original multispectral image to coincide with the corresponding coordinate axes of the regional matrix are counterclockwise rotation. When the ground control point is a rectangular coordinate system, the rotation angles of the two coordinate axes are equal when the original multispectral image is converted to the regional matrix. .

[0011] Furthermore, the affine transformation matrix is ​​used to perform radiation orientation on the ground control points and homonymous points of the original multispectral image, and the errors generated by the overlapping matching of the new multispectral image and the original multispectral image are processed by regional block adjustment to complete the geometric correction of the new multispectral image.

[0012] Furthermore, after the new multispectral image is geometrically corrected to generate pixels in the orthophoto, the original multispectral image is resampled.

[0013] Furthermore, when the coordinate value of the pixel of the orthophoto image in the original multispectral image is an integer, the existing brightness value of the pixel corresponding to the original multispectral image is directly taken out and filled into the new multispectral image; conversely, when the coordinate value of the pixel of the orthophoto image in the original multispectral image is non-integer, the interpolation method is used to accumulate the existing brightness values ​​of the original multispectral image corresponding to the coordinates of other pixels around the pixel whose coordinate values ​​are integers, and the new brightness value of the pixel is obtained by accumulating the existing brightness values ​​of the original multispectral image corresponding to the coordinates of other pixels, and then fill it into the new multispectral image.

[0014] A multispectral imaging device, comprising: Memory for storing computer programs; The processor is used to implement the steps of the multispectral image geometric correction method when executing the computer program.

[0015] A computer-readable storage medium stores a computer program, which implements the steps of a multispectral image geometric correction method when executed by a processor.

[0016] Compared with the prior art, the present invention has the following significant advantages: a geometric correction method is adopted to significantly improve the orthorectification efficiency of multispectral remote sensing images, and a strict geometric data benchmark is provided for the processing of massive multispectral remote sensing data; and, in the process of geometric correction, regional network processing is added to restore the relative and absolute positions between images through overlapping and matching connection points and control points of each image, thereby ensuring the positioning and edge connection accuracy of the images. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of the method for geometric correction of multispectral images according to the present invention. DETAILED DESCRIPTION

[0018] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in combination with the specific content of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the protection scope of the present invention. The contents not described in detail in the embodiments of the present invention belong to the prior art known to professional and technical personnel in this field. The following is a detailed description of the implementation mode of the present invention in combination with the accompanying drawings.

[0019] A method for geometric correction of multispectral images, the method comprising: Extract features from original multispectral images and generate ground control points; Constructing a geometric correction model, in which metadata is stored, and determining the affine transformation relationship between ground control points and metadata; The new multispectral image is input into the model for geometric correction and the orthophoto is output.

[0020] Among them, ground control points are used to meet the needs of geometric correction, affine transformation and registration. They are extracted through field measurement or based on existing geographic location information and serve as a reference for constructing geometric correction models.

[0021] Specifically, the original multispectral image is acquired by remote sensing sensors and is usually stored in image formats (such as JPEG, TIFF, etc.) to generate files. Metadata includes data tags and attribute information of the original multispectral image. The attribute information includes the acquisition date, band information, resolution, geographic location, type of remote sensing sensor, etc. of the original multispectral image.

[0022] Specifically, the gradient value of each pixel in the original multispectral image is calculated, the grayscale change measure of each window centered on each pixel after moving in the surrounding directions is calculated, and each grayscale change measure obtained is compared with the preset empirical threshold to determine the candidate points; the extreme value points are selected from the covariance matrix of the candidate points as the ground control points.

[0023] Specifically, the metadata is preprocessed, including: extracting a geographic reference point in the metadata, creating a first window with a radius of d and the geographic reference point as the center; creating several search windows with a radius of d and the ground control points in the original multispectral image as the center; matching the first window with each search window, wherein the ground control point in the search window with the greatest matching similarity is the point with the same name as the geographic reference point.

[0024] Specifically, the affine transformation relationship between the ground control points and metadata is determined as follows: set the regional matrix, obtain the length plane of the ground control points in the regional matrix, and generate the affine transformation matrix in combination with the length of the ground control points in the original multispectral image; the specific matrix expression is: Where: is the coordinate value of the origin of the original multispectral image in the new coordinate system, are the coordinates of the ground control points in the original multispectral image, are the coordinates of the ground control points in the regional matrix, , is the unit length of the two coordinate axes of the original coordinate system, expressed in the length unit of the corresponding axis of the regional matrix, The scale factor for the selected reference point of the area matrix, , The angles required to rotate the two coordinate axes of the original multispectral image to coincide with the corresponding coordinate axes of the regional matrix are counterclockwise rotation. When the ground control point is a rectangular coordinate system, the rotation angles of the two coordinate axes are equal when the original multispectral image is converted to the regional matrix. .

[0025] Specifically, the affine transformation matrix is ​​used to perform radiation orientation on the ground control points and homonymous points of the original multispectral image, and the errors generated by the overlapping matching of the new multispectral image and the original multispectral image are processed by regional block adjustment to complete the geometric correction of the new multispectral image.

[0026] Conventional geometric correction methods include coarse correction and fine correction. Coarse correction is generally processed by the ground station and only corrects the system error, that is, the original data is geometrically corrected using parameters such as orbit and attitude provided by the satellite, as well as relevant processing parameters in the ground system. Coarse correction is very effective in correcting the internal distortion of remote sensing sensors, but the processed image still has a large residual, so the remote sensing image must be further processed, that is, fine geometric correction. Fine geometric correction is a geometric correction performed using ground control points. It is based on a mathematical model to approximate the geometric distortion process of multispectral images, and uses the control points between the distorted remote sensing image and the standard map to solve the model, and uses this model to correct the geometric distortion. The geometric correction of the present invention refers to fine geometric correction.

[0027] Specifically, after the new multispectral image is geometrically corrected to generate pixels in the orthophoto, the original multispectral image is resampled.

[0028] Specifically, when the coordinate value of the pixel of the orthophoto image in the original multispectral image is an integer, the existing brightness value of the pixel corresponding to the original multispectral image is directly taken out and filled into the new multispectral image; conversely, when the coordinate value of the pixel of the orthophoto image in the original multispectral image is non-integer, the interpolation method is used to obtain the new brightness value of the pixel by accumulating the existing brightness values ​​of the original multispectral image corresponding to the coordinates of other pixels around the pixel whose coordinate values ​​are integers, and then filling it into the new multispectral image.

[0029] A multispectral imaging device, comprising: Memory for storing computer programs; The processor is used to implement the steps of the multispectral image geometric correction method when executing the computer program.

[0030] A computer-readable storage medium stores a computer program, which implements the steps of a multispectral image geometric correction method when executed by a processor.

[0031] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0032] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for geometric correction of multispectral images, characterized by: The method comprises: Extract features from original multispectral images and generate ground control points; Constructing a geometric correction model, in which metadata is stored, and determining the affine transformation relationship between ground control points and metadata; The new multispectral image is input into the model for geometric correction, and an orthophoto is output.

2. The method for geometric correction of multispectral images according to claim 1, characterized in that: The metadata includes data tags and attribute information of the original multispectral image, and the attribute information includes the acquisition date, band information, resolution, and geographic location of the original multispectral image.

3. The method for geometric correction of multispectral images according to claim 2, characterized in that: Calculate the gradient value of each pixel in the original multispectral image, calculate the grayscale change measure after each window centered on each pixel moves in the surrounding directions, and compare each obtained grayscale change measure with a preset empirical threshold to determine the point to be selected; The extreme points are selected from the covariance matrix of the selected points as ground control points.

4. The method for geometric correction based on multispectral images according to claim 3 is characterized in that: The metadata is preprocessed, specifically including: extracting a geographic reference point from the metadata, creating a first window with a radius of d and the geographic reference point as the center; creating a plurality of search windows with a radius of d and the ground control points in the original multispectral image as the center; matching the first window with each search window, wherein the ground control point in the search window with the greatest matching similarity is the point with the same name as the geographic reference point.

5. The method for geometric correction based on multispectral images according to claim 4 is characterized in that: The affine transformation relationship between the ground control points and the metadata is determined as follows: a regional matrix is ​​set, the length plane of the ground control points in the regional matrix is ​​obtained, and the affine transformation matrix is ​​generated in combination with the length of the ground control points in the original multispectral image; the specific matrix expression is: Where: is the coordinate value of the origin of the original multispectral image in the new coordinate system, are the coordinates of the ground control points in the original multispectral image, are the coordinates of the ground control points in the regional matrix, , is the unit length of the two coordinate axes of the original coordinate system, expressed in the length unit of the corresponding axis of the regional matrix, The scale factor for the selected reference point of the area matrix, , The angles required to rotate the two coordinate axes of the original multispectral image to coincide with the corresponding coordinate axes of the regional matrix are counterclockwise rotation. When the ground control point is a rectangular coordinate system, the rotation angles of the two coordinate axes are equal when the original multispectral image is converted to the regional matrix. .

6. The method for geometric correction based on multispectral images according to claim 5 is characterized in that: The affine transformation matrix is ​​used to perform radiation orientation on the ground control points and the same-name points of the original multispectral image, and regional block adjustment processing is performed on the errors generated by the overlapping matching of the new multispectral image and the original multispectral image to complete the geometric correction of the new multispectral image.

7. The method for geometric correction based on multispectral images according to claim 6, characterized in that: After the new multispectral image is geometrically corrected to generate pixels in the orthophoto, the original multispectral image is resampled.

8. The method for geometric correction based on multispectral images according to claim 7 is characterized in that: When the coordinate value of a pixel of the orthophoto image in the original multispectral image is an integer, the existing brightness value of the pixel corresponding to the original multispectral image is directly taken out and filled into the new multispectral image; conversely, when the coordinate value of a pixel of the orthophoto image in the original multispectral image is not an integer, the interpolation method is used to obtain the new brightness value of the pixel by accumulating the existing brightness values ​​of the original multispectral image corresponding to the coordinates of other pixels around the pixel whose values ​​are integers, and then filling the new brightness value into the new multispectral image.

9. A multispectral imaging device, characterized in that: include: Memory for storing computer programs; A processor, used for implementing the steps of the multispectral image geometric correction method as described in any one of claims 1 to 8 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the multispectral image geometric correction method according to any one of claims 1 to 8 are implemented.

Images