A method for radiometric calibration of UAV spectral images using geometric distortion parameters
By establishing a relationship model between geometric distortion parameters and radiation distortion parameters in the spectral image of the drone, the problem of radiation signal distortion in the acquisition of multispectral image data of the drone is solved, high-precision and simple radiation calibration are achieved, and remote sensing quantification is promoted. development of chemical research.
Patent Information
- Application Number
- CN202210471929.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-04-29
- Publication Date
- 2024-11-01
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
During the acquisition process of multi-spectral image data of the UAV, due to factors such as the field of view of the detector, exposure time, and sensor black level, the radiation signal distortion is seriously affected, which seriously affects the acquisition of the final geographic spectrum information. The existing radiation calibration method is complex. And the experimental conditions are high.
Through the drone, the regional series of images of the calibration plate are collected and arranged, the relationship between the image radiation distortion model and the geometric distortion model is analyzed, and the relationship model between geometric distortion parameters and radiation distortion parameters is established, and the radiation calibration is performed using geometric distortion parameters.
It realizes convenient, simple and high-precision drone spectral imaging radiation calibration, reduces experimental conditions and provides a reliable quantitative remote sensing research method.
Smart Images

Figure CN114782552B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of radiometric calibration of multi-spectral remote sensing cameras for unmanned aerial vehicles, and in particular relates to a radiometric calibration method for spectral images of unmanned aerial vehicles using geometric distortion parameters. Background Art
[0002] With the continuous changes in social production methods and the continuous advancement of multispectral sensors, as well as the demand for accurate quantitative inversion of various radiation parameters or physical parameters such as soil moisture content, leaf area, and vegetation index in multiple fields, high-precision quantitative remote sensing using drones equipped with multispectral sensors has become the general trend of the development of drone remote sensing platforms.
[0003] In the process of acquiring multispectral image data from drones, the radiation signal output by the sensor will be distorted due to a series of factors such as the detector field of view, exposure time, sensor black level, extinction effect, etc. This type of radiation distortion will have a serious impact on the final acquisition of ground object spectral information. Therefore, the research on radiometric calibration technology for multispectral data from drones is crucial and is also an important prerequisite for remote sensing quantification. The goal of radiometric calibration is to compensate for the sensor black level, sensor sensitivity, sensor gain and exposure settings, and lens vignetting effects, and to establish a relationship between the radiation brightness received at the sensor entrance pupil and the DN value output by the sensor. Traditional radiometric calibration methods require a large amount of prior data related to the camera and field test conditions, and then calculate the results through complex mathematical formulas. This method has high requirements for experimental conditions and a complex calculation process.
[0004] The geometric distortion correction technology of remote sensing cameras has been developed to maturity. The geometric distortion parameters of the camera can be obtained through relatively simple experiments, and the experimental accuracy of traditional methods can reach a high level. Summary of the invention
[0005] Aiming at the radiation calibration process of UAV multispectral remote sensing, the present invention proposes a method for UAV spectral image radiation calibration using geometric distortion parameters. A series of images of the area where the calibration plate is arranged are collected by UAV, the relationship between the image radiation distortion model and the geometric distortion model is analyzed, and a relationship model between the geometric distortion parameters and the radiation distortion parameters is established, so as to realize reliable UAV spectral image radiation calibration based on geometric distortion parameters.
[0006] To achieve the above object, the present invention provides a method for calibrating the radiometric quality of UAV spectral images using geometric distortion parameters, comprising the following steps:
[0007] Step 1) UAV spectral image data collection, including using a multi-spectral camera mounted on a UAV to take multi-directional and multi-altitude photos of the deployed reflectors;
[0008] Step 2) UAV spectral image preprocessing, including extracting the reflector area of the original image taken by the UAV and reading the DN value of the multispectral image;
[0009] Step 3) calibrating the geometric distortion parameters of the UAV spectral camera, including selecting a camera geometric calibration method, solving a camera geometric distortion model to obtain camera geometric distortion parameters;
[0010] Step 4) establishing the relationship between the geometric distortion model and the radiation distortion model, including establishing a mathematical model for the relationship based on the fact that both the DN value correction value and the geometric distortion correction value are correlated with the coordinate position of the reflector in the image, assuming that the DN value correction value and the geometric distortion correction value of pixels at the same position are correlated;
[0011] Step 5) radiation distortion correction, including performing radiation calibration using image geometric distortion parameters based on the relationship model between geometric distortion and radiation distortion obtained in step 4).
[0012] Moreover, the drone spectral image data collection in step 1) is implemented as follows:
[0013] Step 1.1, select the shooting area and arrange reflectors with different reflectivity;
[0014] Step 1.2, configure the multispectral camera parameters carried by the drone, including the camera exposure time and the multispectral band;
[0015] Step 1.3, plan the flight route of the drone according to the terrain and topography of the flight area;
[0016] Step 1.4, collect multispectral image data at different locations and altitudes by flying the drone, and the camera shooting area accommodates all the deployed reflectors.
[0017] Moreover, the UAV spectral image preprocessing in step 2) is implemented as follows:
[0018] Step 2.1, sorting all captured image data by band classification;
[0019] Step 2.2, read the pixel DN value of each reflector area in the image, calculate the regional mean, and obtain the DN value corresponding to each reflectivity of the image;
[0020] Step 2.3, record the position of each reflector in each image;
[0021] Step 2.4, calculate the DN value corresponding to each reflectivity according to the factory parameters of the camera, compare the value with the value calculated in step 2.2, and obtain the DN value correction value.
[0022] Moreover, the geometric distortion parameter calibration of the UAV spectral camera in step 3) is implemented as follows:
[0023] Step 3.1, set up the UAV multispectral camera geometric calibration plate;
[0024] Step 3.2, operate the multispectral camera of the UAV to collect the image data of the calibration plate;
[0025] Step 3.3, extract the checkerboard corner points in the image, establish the camera geometric distortion model according to the Zhang Zhengyou calibration method, and calculate the camera geometric distortion parameters.
[0026] Moreover, the relationship between the geometric distortion model and the radiation distortion model in step 4) is established, and the implementation method is as follows:
[0027] Step 4.1, using the geometric distortion parameters obtained in step 3, calculate the geometric distortion correction value of the location of the calibration plate in each image;
[0028] Step 4.2, analyzing the relationship between the DN value correction value and the coordinate position of the reflector in the image, taking the position of the reflector in the image as input and the DN value correction value at the position as output, and fitting a quartic polynomial model using the linear least squares method;
[0029] Step 4.3, analyzing the relationship between the geometric distortion correction value and the coordinate position of the reflector in the image, and fitting the mathematical model;
[0030] Step 4.4, using the results of step 4.2 and step 4.3, establish a relationship model between the geometric distortion model and the radiation distortion model.
[0031] Furthermore, step 5) radiation distortion correction is implemented as follows,
[0032] Step 5.1, traverse the remote sensing images of each spectral band, use the geometric distortion and radiation distortion relationship model corresponding to the band and the height, and directly use the geometric distortion parameters to calibrate the radiation of each image;
[0033] Step 5.2, add the original image and the calculated DN value correction value to obtain the radiometrically calibrated image and output it.
[0034] The present invention has the following positive effects:
[0035] 1) The present invention proposes a method for calibrating the radiation of UAV spectral images using geometric distortion parameters, which provides a convenient, fast and reliable radiation calibration method for quantitative remote sensing, which has been developed and applied in various fields in recent years, and for remote sensing fine inversion related research.
[0036] 2) The present invention proposes for the first time that there is a correlation between geometric distortion and radiation distortion in multispectral remote sensing, which provides a new idea for quantitative research on remote sensing and is of great significance to the application of remote sensing in various fields.
[0037] 3) The radiation calibration method proposed in the present invention solves the problem of high experimental conditions and many prior data requirements in traditional radiation calibration. The method is easy to operate, simple to calculate and has high calibration accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] Figure 1 The figure is a flow chart of a method according to an embodiment of the present invention. DETAILED DESCRIPTION
[0039] The technical solution of the present invention is described in detail below with reference to the accompanying drawings and embodiments.
[0040] like Figure 1 As shown, the embodiment provides a method for calibrating the radiometric quality of a UAV spectral image using geometric distortion parameters, comprising the following steps:
[0041] Step 1) UAV spectral image data collection: This step mainly uses a multi-spectral camera mounted on a UAV to take multi-directional and multi-altitude photos of the deployed reflectors.
[0042] In specific implementation, the preferred data collection method is as follows:
[0043] First, on a clear afternoon, select an area with a wide field of vision and place reflectors with different reflectivity in the area. The specific implementation requirements of this process are:
[0044] ① Select reflectors of the same size, material and produced in the same batch.
[0045] ② The reflectivity of the reflective panels should be evenly distributed. The preferred method is to use 7 reflective panels at 5%, 10%, 20%, 30%, 50%, 70% and 90%.
[0046] ③The reflectors should not be placed too far apart, and the distance between adjacent reflectors should be controlled between 30-50cm.
[0047] Secondly, configure the shooting parameters of the drone's multispectral camera. The recommended implementation options for this process are:
[0048] ① Manually set the camera’s exposure time to 1 / 8000 second.
[0049] ②The camera bands are set to: blue band (450nm), green band (550nm), red band (650nm), red edge band (725nm), and near infrared band (850nm).
[0050] Thirdly, be familiar with the terrain around the selected deployment area, plan the flight route of the drone, and ensure that the reflector is fully captured and appears at various positions in the spectral image.
[0051] Finally, the drone is manually flown to collect multispectral image data at different altitudes.
[0052] Step 2) UAV spectral image preprocessing. This step mainly involves preprocessing operations such as reflector area extraction and multispectral image DN value reading on the original images taken by the drone, in preparation for the subsequent analysis of relationships and the establishment of mathematical models.
[0053] First, all the images captured are sorted by bands, and separate folders are created for different bands. The images captured in the same time and corresponding to each band are named one by one to facilitate comparative analysis.
[0054] Secondly, open an image, extract the ROI area (region of interest) corresponding to the reflector in the image, read the DN value of the pixels in the area, calculate the average DN value of all pixels, and use this value as the corresponding DN value of the reflectivity in the image;
[0055] At the same time, take the center point of the reflector as the location and record the coordinates of each reflector in the image. Repeat the above operation for each image and organize the results into a table.
[0056] Finally, find the relevant production information (camera factory parameters) of the multispectral camera used in the experiment, and calculate the DN value (brightness value) corresponding to each reflectivity captured by the camera. Compare the DN value corresponding to the reflectivity in each image obtained in the previous step with this value, calculate the DN value correction value, and organize the results into a table.
[0057] Step 3) Calibration of geometric distortion parameters of the UAV spectral camera. This step mainly involves selecting a camera geometric calibration method and solving the camera geometric distortion model to obtain the camera geometric distortion parameters.
[0058] In specific implementation, the preferred data collection method is as follows:
[0059] First, the classic Zhang Zhengyou chessboard calibration method is used to select a chessboard calibration plate of appropriate size according to the geometric resolution of the camera.
[0060] Secondly, choose an open indoor environment to place the calibration plate, manually operate the drone, and take 15-25 images of the checkerboard calibration plate from various angles.
[0061] Finally, the checkerboard corner points in the image are extracted, and the camera geometric distortion model is established according to the principle of Zhang Zhengyou calibration method, and the camera geometric distortion parameters are calculated.
[0062] Zhang Zhengyou's calibration method only considers the radial distortion, which has a greater impact in the distortion model, and solves for the radial distortion parameter k 1 , k 2 , the camera geometric distortion model formula is shown in formula (1).
[0063]
[0064] Where: x, y are the coordinates of the image plane where the pixel is located. are the coordinates after geometric distortion correction.
[0065] Step 4) The relationship between the geometric distortion model and the radiation distortion model is established. This step is based on the previous three steps. The DN value correction value and the geometric distortion correction value are both correlated with the coordinate position of the reflector in the image, that is, the DN value correction value and the geometric distortion correction value of the pixel at the same position are correlated. A mathematical model is established for this relationship.
[0066] First, using the geometric distortion model and geometric distortion parameters obtained in step 3, geometric distortion correction is performed band by band and image by image. The corrected image is compared with the original image, and the geometric distortion correction value of the pixel at the location of the reflector is calculated and organized into a table.
[0067] Secondly, visualize the DN value correction table obtained in step 2 and analyze the relationship between the data, that is, the relationship between the DN value correction and the coordinate position of the reflector in the image. Take the position of a pixel in the multispectral image (the image where the reflector is located) as input and the DN value correction at that position as output. Use the linear least squares method to fit the quartic polynomial model. The implementation requirements of this process are:
[0068] ① The image plane coordinate system of the image taken by the UAV multispectral camera is used to represent the position of the pixel in the image.
[0069] ② After determining the image plane coordinate position of a certain pixel point, use the coordinates to calculate the distance from the point to the origin, which is used as the input for model fitting.
[0070] ③The determination coefficient of the fitted quartic polynomial model is greater than 0.9.
[0071] In this embodiment, the fitted quartic polynomial is as follows, where p 1 、p 2 、p 3 、p 4 、p 5 The coefficients of the fitted polynomials are given with a 95% confidence interval.
[0072]
[0073]
[0074] Where f(r) represents the DN value correction value.
[0075] Finally, there is a definite one-to-one correspondence between the geometric distortion correction value and the pixel location. A mathematical model is established with the geometric distortion correction value as input and the DN value correction value of the corresponding pixel as output. This model is calculated by the geometric distortion model selected in step 3. Among them, Δx and Δy are the geometric distortion correction values of the image plane coordinates where the reflector is located.
[0076]
[0077] Step 5), radiation distortion correction, this step is based on the relationship model between geometric distortion and radiation distortion obtained in the above steps, and uses the image geometric distortion parameters to perform radiation calibration.
[0078] Firstly, the remote sensing images of each spectral band are traversed, and the geometric distortion and radiation distortion relationship model corresponding to the height of the band is used to directly calibrate the radiation of each image with the geometric distortion parameters.
[0079] Finally, the original image is added to the calculated DN value correction value to obtain the radiometrically calibrated image, and the resulting image after radiometric calibration using the method proposed in this patent is output.
[0080] In specific implementation, the method proposed in the technical solution of the present invention can be implemented by technical personnel in this field using computer software technology to realize automatic operation process. System devices for implementing the method, such as computer-readable storage media storing the corresponding computer program of the technical solution of the present invention and computer equipment running the corresponding computer program, should also be within the scope of protection of the present invention.
[0081] In some possible embodiments, a system for calibrating UAV spectral image radiation using geometric distortion parameters is provided, including a processor and a memory, the memory being used to store program instructions, and the processor being used to call the stored instructions in the memory to execute a method for calibrating UAV spectral image radiation using geometric distortion parameters as described above.
[0082] In some possible embodiments, a system for calibrating UAV spectral image radiation using geometric distortion parameters is provided, including a readable storage medium, on which a computer program is stored. When the computer program is executed, a method for calibrating UAV spectral image radiation using geometric distortion parameters as described above is implemented.
[0083] The specific embodiments described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.
Claims
1. A method for calibrating the radiometric quality of unmanned aerial vehicle spectral images using geometric distortion parameters, characterized in that: The steps include: Step 1) UAV spectral image data collection, including using a multi-spectral camera mounted on a UAV to take multi-directional and multi-altitude photos of the deployed reflectors; Step 2) UAV spectral image preprocessing, including extracting the reflector area of the original image taken by the UAV and reading the DN value of the multispectral image; Step 3) calibrating the geometric distortion parameters of the UAV spectral camera, including selecting a camera geometric calibration method, solving a camera geometric distortion model to obtain camera geometric distortion parameters; Step 4) The relationship between the geometric distortion model and the radiation distortion model is established, including that the DN value correction value and the geometric distortion correction value are both correlated with the coordinate position of the reflector in the image, and it is considered that the DN value correction value and the geometric distortion correction value of the pixel at the same position are correlated, and a mathematical model is established for this relationship; the implementation method is as follows: Step 4.1, using the geometric distortion parameters obtained in step 3, calculate the geometric distortion correction value of the location of the calibration plate in each image; Step 4.2, analyzing the relationship between the DN value correction value and the coordinate position of the reflector in the image, taking the position of the reflector in the image as input and the DN value correction value at the position as output, and fitting a quartic polynomial model using the linear least squares method; Step 4.3, analyzing the relationship between the geometric distortion correction value and the coordinate position of the reflector in the image, and fitting the mathematical model; Step 4.4, using the results of step 4.2 and step 4.3, establish a relationship model between the geometric distortion model and the radiation distortion model; Step 5) radiation distortion correction, including performing radiation calibration using image geometric distortion parameters based on the relationship model between geometric distortion and radiation distortion obtained in step 4); The implementation is as follows, Step 5.1, traverse the remote sensing images of each spectral band, use the geometric distortion and radiation distortion relationship model corresponding to different heights of the band, and directly use the geometric distortion parameters to calibrate the radiation of each image; Step 5.2, add the original image and the calculated DN value correction value to obtain the radiometrically calibrated image and output it.
2. The method for calibrating the radiometric quality of unmanned aerial vehicle spectral images using geometric distortion parameters according to claim 1, characterized in that: The UAV spectral image data collection in step 1) is implemented as follows: Step 1.1, select the shooting area and arrange reflectors with different reflectivity; Step 1.2, configure the multispectral camera parameters carried by the drone, including the camera exposure time and the multispectral band; Step 1.3, plan the flight route of the drone according to the terrain and topography of the flight area; Step 1.4, collect multispectral image data at different locations and altitudes by flying the drone, and the camera shooting area accommodates all the deployed reflectors.
3. The method for calibrating the radiometric quality of unmanned aerial vehicle spectral images using geometric distortion parameters according to claim 1, characterized in that: The preprocessing of the drone spectral image in step 2) is implemented as follows: Step 2.1, sorting all captured image data by band classification; Step 2.2, read the pixel DN value of each reflector area in the image, calculate the regional mean, and obtain the DN value corresponding to each reflectivity of the image; Step 2.3, record the position of each reflector in each image; Step 2.4, calculate the DN value corresponding to each reflectivity according to the factory parameters of the camera, compare the value with the value calculated in step 2.2, and obtain the DN value correction value.
4. The method for calibrating the radiometric quality of unmanned aerial vehicle spectral images using geometric distortion parameters according to claim 1, characterized in that: The geometric distortion parameter calibration of the UAV spectral camera in step 3) is implemented as follows: Step 3.1, set up the UAV multispectral camera geometric calibration plate; Step 3.2, operate the multispectral camera of the UAV to collect the image data of the calibration plate; Step 3.3, extract the checkerboard corner points in the image, establish the camera geometric distortion model according to the Zhang Zhengyou calibration method, and calculate the camera geometric distortion parameters.
