Indoor and outdoor integrated calibration method, system, electronic equipment and storage medium

Abstract

The present invention discloses an integrated indoor and outdoor calibration method, system, electronic device, and storage medium. The method comprises: collecting indoor calibration data, calculating the indoor calibration data, and establishing an initial calibration model; wherein the initial calibration model includes a geometric calibration model and / or a radiation calibration model; collecting outdoor verification data, inputting the outdoor verification data into the initial calibration model, and obtaining a verification result; wherein the outdoor verification data and the indoor calibration data have the same data type; and if the verification result is less than or equal to a preset error, outputting the initial calibration model, which is used to calibrate the image. The present invention fully considers the impact of indoor and outdoor environmental data on images captured by a multispectral camera and proposes a complete integrated indoor and outdoor calibration method.

Description

Technical Field

[0001] The present invention relates to the field of remote sensing technology, and in particular to an indoor and outdoor integrated calibration method, system, electronic equipment and storage medium. Background Art

[0002] As a novel Earth observation technology, drone-mounted low-altitude multispectral cameras can acquire the multispectral characteristics, texture features, and accurate geospatial location of target objects, enabling quantitative inversion of important physical, chemical, and biological parameters of these objects. Calibration of the multispectral camera's geometric, radiometric, and spectral performance is a crucial prerequisite for highly reliable and accurate airborne low-altitude multispectral perception. However, because airborne multispectral cameras integrate a novel multispectral detection system and operating mechanism, their response characteristics are complex, requiring a significant amount of calibration data. Furthermore, due to the complex atmospheric variations in outdoor low-altitude environments, multispectral camera calibration is also affected by environmental factors such as atmospheric radiation. Therefore, calibration methods used for traditional satellite and space-based remote sensing sensors are not applicable to multispectral cameras. The existing technology lacks a comprehensive indoor and outdoor calibration method for drone-mounted multispectral cameras, and the construction of a calibration site for multispectral cameras is also unclear. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to overcome the defect in the prior art of lacking a complete indoor and outdoor calibration method for multispectral cameras on drones, and to provide an indoor and outdoor integrated calibration method, system, electronic equipment and storage medium.

[0004] The present invention solves the above technical problems through the following technical solutions:

[0005] In one aspect, the present invention provides an indoor-outdoor integrated calibration method, which is applied to the calibration of images captured by a multispectral camera. The indoor-outdoor integrated calibration method includes:

[0006] Collecting indoor calibration data, calculating the indoor calibration data and establishing an initial calibration model; wherein the initial calibration model includes a geometric calibration model and / or a radiation calibration model;

[0007] Collecting outdoor verification data, inputting the outdoor verification data into the initial calibration model, and obtaining a verification result; wherein the data type of the outdoor verification data is the same as that of the indoor calibration data;

[0008] If the verification result is less than or equal to the preset error, the initial correction model is output, and the initial correction model is used to calibrate the image.

[0009] Preferably, the steps of collecting indoor calibration data, calculating the indoor calibration data and establishing an initial calibration model include:

[0010] Collect indoor spectrum calibration data to determine the spectrum environment;

[0011] Under the same spectral environment, indoor geometric calibration data is collected, and the corner points of the indoor geometric calibration data are calculated according to the checkerboard calibration method. The distortion coefficient is estimated using the least squares method, and then the maximum likelihood method is used to optimize the estimation accuracy. A geometric correction model is established based on the corner points, the distortion coefficient and the estimation accuracy.

[0012] Preferably, the steps of collecting indoor calibration data, calculating the indoor calibration data and establishing an initial calibration model include:

[0013] Collecting indoor radiation calibration data, inputting the indoor radiation calibration data into the geometric correction model, and outputting corrected indoor radiation calibration data;

[0014] The corrected indoor radiation calibration data is processed with dark current calibration, vignetting effect calibration and quantum efficiency fitting, and a radiation correction model is established according to the processing results.

[0015] Preferably, the steps of collecting outdoor verification data, inputting the outdoor verification data into the initial calibration model, and obtaining the verification results include:

[0016] If the verification result is greater than the preset error, outdoor calibration data is collected, and the initial calibration model is corrected according to the indoor calibration data and the outdoor calibration data, where the data types of the outdoor calibration data and the indoor calibration data are the same.

[0017] Preferably, the indoor and outdoor integrated calibration method further includes:

[0018] Collecting outdoor atmospheric calibration data, inputting the outdoor atmospheric calibration data into the geometric correction model and the radiation correction model, and outputting corrected outdoor atmospheric calibration data;

[0019] Determine the atmospheric path radiation value based on the radiance value of the zero reflectivity target;

[0020] An atmospheric correction model is established according to the atmospheric radiation value and the outdoor atmospheric calibration data, and the atmospheric correction model is used to calibrate the image.

[0021] Preferably, the indoor and outdoor integrated calibration method further includes:

[0022] Collecting outdoor reflectivity calibration data, and sequentially inputting the outdoor reflectivity calibration data into the geometric correction model, the radiation correction model, and the atmospheric correction model, and outputting corrected outdoor reflectivity calibration data;

[0023] The dark target method and the empirical linear method are used to obtain the downlink radiance value corresponding to the atmospheric path radiation value; a reflectance correction model is established based on the downlink radiance value and the outdoor reflectance calibration data, and the reflectance correction model is used to calibrate the image.

[0024] Preferably, the indoor and outdoor integrated calibration method further includes:

[0025] The calibration data of the image is input into the initial calibration model, and an image that meets the preset calibration data is output. The calibration data of the image has the same data type as the indoor calibration data.

[0026] Preferably, the step of collecting outdoor verification data includes:

[0027] Collect outdoor verification data at a portable outdoor calibration site;

[0028] The equipment of the movable outdoor calibration field includes at least one of a movable standard diffuse reflection plate target, a movable resolution target, a movable geometric target and a movable zero reflectivity target;

[0029] The movable zero-reflectivity target is a hollow cube with a surface painted with black matte paint, a hole is set on the top of the cube, and the interior of the cube is filled with black diffuse reflection cloth.

[0030] Preferably, the step of collecting indoor calibration data includes:

[0031] Collect indoor calibration data at the indoor calibration site;

[0032] The equipment in the indoor calibration field includes a monochromator, a standard radiation source and a geometric target;

[0033] Wherein, the standard radiation source includes a standard diffuse reflection plate and / or a standard integrating sphere;

[0034] The monochromator and the standard radiation source are arranged on an optical platform and in a dark room.

[0035] Preferably, the indoor and outdoor integrated calibration method further includes:

[0036] The following formula is used to perform the vignetting effect correction:

[0037] VE=ax 3 +by 3 +cyx 2 +dxy 2 +ex 2 +fy 2 +gxy+hx+iy+j

[0038] Among them, VE is the correction coefficient output by the model, a to j are fitting coefficients, and x and y are the position coordinates of all pixels in the image.

[0039] Preferably, the indoor and outdoor integrated calibration method further includes:

[0040] The quantum efficiency fitting is performed using the following formula:

[0041]

[0042] Where L(i0) is the radiance value of the image principal point, DN(i0) is the pixel value of the image principal point, and k, m, and n are fitting coefficients.

[0043] Preferably, the indoor and outdoor integrated calibration method further includes:

[0044] The atmospheric correction model is fitted using the following formula:

[0045]

[0046] Among them, L0 is the atmospheric radiation value, L p is the downlink radiance value, e is a natural constant, r and s are fitting coefficients.

[0047] In a second aspect, the present invention provides an indoor and outdoor integrated calibration system for calibrating images captured by a multispectral camera. The indoor and outdoor integrated calibration system includes:

[0048] an acquisition module, configured to acquire indoor calibration data, calculate the indoor calibration data, and establish an initial calibration model; wherein the initial calibration model includes a geometric calibration model and / or a radiation calibration model; and further configured to acquire outdoor verification data, input the outdoor verification data into the initial calibration model, and obtain a verification result; wherein the outdoor verification data is of the same data type as the indoor calibration data;

[0049] The verification module is configured to output the initial correction model if the verification result is less than or equal to a preset error, wherein the initial correction model is used to calibrate the image.

[0050] In a third aspect, the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned indoor and outdoor integrated calibration method when executing the computer program.

[0051] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-mentioned indoor and outdoor integrated calibration method.

[0052] The positive progress effect of the present invention is:

[0053] This method collects indoor verification data to establish an initial calibration model. This model then inputs collected outdoor verification data to generate a verification result. If the verification result is less than or equal to a preset error, the initial calibration model is output and applied to the calibration image. This method fully considers the impact of indoor and outdoor environmental data on multispectral camera images, proposing a comprehensive, integrated indoor and outdoor calibration method. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a flow chart of the indoor and outdoor integrated calibration method according to Example 1 of the present invention;

[0055] Figure 2 This is a flow chart of establishing a geometric correction model according to embodiment 1 of the present invention;

[0056] Figure 3 This is a flow chart of establishing a radiation correction model according to Example 1 of the present invention;

[0057] Figure 4 This is a flow chart of establishing an atmospheric correction model according to Example 1 of the present invention;

[0058] Figure 5 This is a flow chart of establishing a reflectivity correction model according to Example 1 of the present invention;

[0059] Figure 6 This is a calibration flow chart of the narrow-band multispectral camera in the 675nm band according to Example 1 of the present invention;

[0060] Figure 7 This is a diagram showing the target arrangement of a movable outdoor calibration field according to Example 1 of the present invention;

[0061] Figure 8 The route map obtained from the outdoor calibration data of Example 1 of the present invention;

[0062] Figure 9 This is a diagram showing the spectrum calibration and fitting results of the narrow-band multispectral camera in Example 1 of the present invention in the 675nm band;

[0063] Figure 10 This is a dark current calibration flow chart of the narrow-band multispectral camera in the 675nm band according to Example 1 of the present invention;

[0064] Figure 11 This is a diagram of correction coefficients for vignetting effect calibration using a lookup table method in the 675nm band for the narrow-band multispectral camera of Example 1 of the present invention;

[0065] Figure 12a This is an example of an image of the narrow-band multispectral camera of Example 1 of the present invention before vignetting effect correction in the 675nm band;

[0066] Figure 12b This is an example of an image after vignetting effect correction is performed on the narrow-band multispectral camera of Example 1 of the present invention in the 675nm band;

[0067] Figure 13 This is a graph showing the quantum efficiency fitting results of the narrow-band multispectral camera of Example 1 of the present invention in the 675nm band using the exponential fitting method;

[0068] Figure 14 This is a graph showing the atmospheric correction fitting results of the narrow-band multispectral camera in Example 1 of the present invention in the 675nm band;

[0069] Figure 15a This is an image example of the narrow-band multispectral camera of Example 1 of the present invention before indoor and outdoor integrated calibration in the 675nm band;

[0070] Figure 15b This is an image example of the narrow-band multispectral camera of Example 1 of the present invention after indoor and outdoor integrated calibration in the 675nm band;

[0071] Figure 16 This is a structural diagram of an indoor and outdoor integrated calibration system according to Example 2 of the present invention;

[0072] Figure 17 This is a schematic structural diagram of an electronic device according to embodiment 3 of the present invention. DETAILED DESCRIPTION

[0073] The present invention is further described below by way of examples, but the present invention is not limited to the scope of the examples.

[0074] Example 1

[0075] This embodiment provides an indoor and outdoor integrated calibration method, which is to calibrate the low-altitude multispectral camera on the UAV and calibrate the images taken by the multispectral camera. Figure 1 , the indoor and outdoor integrated calibration method includes:

[0076] S101: Collect indoor calibration data, calculate indoor calibration data and establish an initial calibration model.

[0077] Wherein, the initial correction model includes a geometric correction model and / or a radiation correction model;

[0078] S102: Collect outdoor verification data, input the outdoor verification data into the initial calibration model, and obtain a verification result.

[0079] The data types of the outdoor verification data and the indoor calibration data are the same.

[0080] S103: If the verification result is less than or equal to the preset error, the initial correction model is output.

[0081] Among them, the preset error is set according to the actual situation, and the initial correction model is used to calibrate the image.

[0082] In addition, if the exposure time, white balance, filter, lens, or photosensitive element of the multispectral camera has been adjusted, the above steps must be repeated for recalibration.

[0083] This embodiment fully considers the impact of indoor and outdoor environmental data on the images captured by the multispectral camera, and proposes a complete indoor and outdoor integrated calibration method. When the calibration data collected indoors has a high accuracy, an initial calibration model is established, and the outdoor verification data is input into the initial calibration model for verification. In this way, the impact of outdoor environmental data on the image is taken into account at the same time, the accuracy of the initial calibration model is improved, and the accuracy of the calibration result is further improved.

[0084] In one embodiment, see Figure 2 , step S101 includes:

[0085] S1011. Collect indoor spectrum calibration data and determine the spectrum environment;

[0086] Among them, indoor spectral calibration data includes monochromator image data with continuous band changes, etc.

[0087] Collecting indoor spectral calibration data can determine the spectral environment of the multispectral camera, such as the spectral range and number of bands. The spectral environment of the multispectral camera is fixed, so it is necessary to establish an initial calibration model based on the determination of the spectral environment.

[0088] Specifically, based on the indoor spectral calibration dataset, the average value of 3×3 pixel values ​​at the pixel center position of the monochromator output light source in the image captured by the multispectral camera is extracted. The Gaussian fitting method is used to fit the relationship between the pixel value and the monochromator output wavelength. The highest point of the curve is taken as the central wavelength of the multispectral camera, and the half-maximum width of the curve is taken as the bandwidth.

[0089] S1012. Under the same spectral environment, collect indoor geometric calibration data, calculate the corner points of the indoor geometric calibration data based on the checkerboard calibration method, apply the least squares method to estimate the distortion coefficient, and then use the maximum likelihood method to optimize the estimation accuracy. Establish a geometric correction model based on the corner points, distortion coefficient and estimation accuracy.

[0090] Among them, indoor geometric calibration data mainly includes image data of geometric targets at different distances and angles.

[0091] Specifically, before collecting indoor geometry calibration data, the camera exposure time is fixed. Then, by adjusting the orientation of the indoor geometry target or the camera, a certain number of photos of the geometric target in different orientations are taken to obtain an indoor geometry calibration dataset. Based on the indoor geometry calibration dataset and the checkerboard calibration method, the geometric target corners are extracted from the photos. Under the ideal distortion-free condition, the distortion coefficients under actual radial distortion are estimated using the least squares method, combining the intrinsic parameters of five multispectral cameras and the extrinsic parameters of six multispectral cameras. Finally, the maximum likelihood method is used to optimize the estimation, improve the estimation accuracy, establish a geometric correction model, and complete the geometric correction.

[0092] In this embodiment, the spectral environment is determined through indoor spectral calibration data, and a geometric calibration model is established through indoor geometric calibration data to complete geometric calibration, thereby improving the geometric accuracy of images captured by the multispectral camera.

[0093] In one embodiment, see Figure 3 , step S101 further includes:

[0094] S1013. Collect indoor radiation calibration data, input the indoor radiation calibration data into a geometric correction model, and output the corrected indoor radiation calibration data.

[0095] Among them, the indoor radiation calibration data mainly includes dark current calibration images taken in a dark environment without light; vignetting effect calibration images of standard radiation sources that output parallel and uniform light taken under different radiance conditions; and quantum efficiency fitting images of standard radiation sources taken under continuously changing radiance conditions.

[0096] The output corrected indoor radiation calibration data is an image without geometric distortion.

[0097] Specifically, indoor radiation calibration data consists of three main parts: data required for dark current calibration, data required for vignetting effect calibration, and data required for quantum efficiency fitting. By adjusting the monochromator's output wavelength in increments of 1 nanometer, from a value where the multispectral camera has almost no pixel response, to a value where the pixel responds at maximum, and then to a value where it responds at almost no pixel response, a single image is captured each time the monochromator's output wavelength is adjusted.

[0098] The data required for dark current calibration is collected by using a multispectral camera with the lens cap closed and no light shining on it. The camera captures several images at different ISO (sensitivities) or gain levels. For example, five images can be taken at each ISO or gain level.

[0099] The data collection method for vignetting calibration is to select several standard radiation source modes with different radiances. The light output from the standard radiation source must be uniform, that is, the radiance at different output positions is the same and the light is parallel. Set the camera to automatic ISO or gain mode and capture several images under each radiance condition. For example, select five radiance modes and capture five images under each radiance condition.

[0100] The data required for quantum efficiency fitting is collected by adjusting the radiance of the standard radiation source at different ISO or gain modes of the multispectral camera, so that the multispectral camera's response ranges from nearly underexposure to nearly overexposure. Several images are captured using the multispectral camera with different responses, while the corresponding current or voltage value of the standard radiation source is recorded. Each ISO or gain mode contains a set of images ranging from nearly underexposure to nearly overexposure. Because the standard radiation source is precalibrated, the radiance value for any output wavelength can be obtained based on the calibration results and the current or voltage value. For example, ten images are captured at different radiance conditions at different ISO or gain modes.

[0101] S1014. Perform dark current calibration, vignetting effect calibration, and quantum efficiency fitting on the corrected indoor radiation calibration data, and establish a radiation calibration model based on the processing results.

[0102] Specifically, the collected indoor radiation calibration data is input into the geometric correction model, and then the corrected indoor radiation calibration data is subjected to radiation calibration. The radiation calibration processing mainly includes: dark current calibration, vignetting effect calibration and quantum efficiency fitting.

[0103] A lookup table (LUT) is used for dark current calibration. Considering the spatial non-uniformity of two-dimensional images, a template with the same two-dimensional structure as the image is used as the dark current lookup table. Because dark current is affected by ISO or gain, different dark current lookup tables are required for different ISO or gain modes. In the data set required for indoor dark current calibration, each ISO or gain mode has several images. The absolute value of the mean pixel value of all images is taken, and each value is filled into the template at the position corresponding to the pixel in the image. When correcting for dark current, the value of the corresponding position in the template should be subtracted from all pixels in the image to be corrected.

[0104] The following method is used for vignetting effect calibration: first, Gaussian filtering is performed on the data set required for indoor vignetting effect calibration to reduce noise, and then the vignetting effect correction coefficient is calculated using the following formula:

[0105]

[0106] Among them, VE factoris the correction coefficient, x, y are the position coordinates of the pixel in the image, V(x, y) is the pixel value at all positions in the image, is the pixel value of the principal point.

[0107] Secondly, a lookup table (LUT) method is used. This method takes the average value of the correction coefficient for each pixel position of all images. Similarly, a template with the same two-dimensional structure as the image is established based on the average value of each pixel. When correcting the vignetting effect, the pixel value of each position of the image to be corrected is multiplied by the corresponding correction coefficient in the template.

[0108] For quantum efficiency fitting, the radiance value corresponding to each image in the data set required for quantum efficiency fitting is obtained based on the center wavelength of the multispectral camera and the recorded current and voltage values ​​of the standard radiation source. A linear fitting method is used to establish a linear fitting relationship between image pixel values ​​and radiance values. Each ISO or gain mode requires a corresponding fitting model. The fitting formula is as follows:

[0109] L(i0)=k1×DN(i0)+m1

[0110] Where L(i0) is the radiance value of the image principal point, DN(i0) is the pixel value of the image principal point, and k1 and m1 are fitting coefficients.

[0111] In this embodiment, radiation correction is completed by processing indoor radiation calibration data, thereby improving the radiation accuracy of images captured by the multispectral camera.

[0112] In one embodiment, step S102 includes:

[0113] If the verification result is greater than the preset error, outdoor calibration data is collected, and the initial calibration model is corrected based on the indoor calibration data and the outdoor calibration data, and the data types of the outdoor calibration data and the indoor calibration data are the same.

[0114] Specifically, for the correction of the geometric correction model: based on the geometric target images taken at different heights, different positions and different angles in the outdoor geometric calibration data set, a part of the images are allocated as outdoor geometric verification data, and the other part of the images are used as outdoor geometric calibration data. The reprojection error (i.e., verification result) of the cross center position of the geometric target is calculated. When the error is less than or equal to the preset reprojection error (i.e., preset error), it means that the geometric correction model can be used directly; if the error is greater than the preset reprojection error, it is necessary to combine the indoor geometric calibration data and the outdoor geometric calibration data, calibrate them through the checkerboard calibration method, correct the geometric correction model, and obtain an integrated indoor and outdoor geometric correction model. Among them, the preset reprojection error can be 1 pixel.

[0115] To correct the radiation correction model, a portion of the outdoor ground-collected radiation calibration data set is divided into outdoor radiation verification data and another portion is used as outdoor radiation calibration data. The outdoor radiation verification data is input into the geometric correction model, ultimately obtaining the radiance value of the outdoor radiation verification data image. The absolute percentage error (i.e., the verification result) between the image radiance and the true radiance measured by the ground object spectrometer is calculated. If the error is within the preset absolute percentage error (i.e., the preset error), the radiation correction model can be directly adopted. If it is higher than the preset error, a new quantum efficiency fitting model is obtained by combining the outdoor radiation calibration data with the quantum efficiency of the dataset required for indoor quantum efficiency fitting. The radiation correction model is then corrected by combining the new quantum efficiency fitting results with the original dark current calibration results and the original vignetting effect calibration results. The preset absolute percentage error is 10%.

[0116] In this embodiment, if the verification result is greater than the preset error, it indicates that outdoor environmental factors will greatly affect the calibration of the multispectral camera. Therefore, it is necessary to combine indoor and outdoor calibration data to correct the initial calibration model to improve the calibration accuracy of the calibration model.

[0117] In practice, in order to calibrate a multispectral camera in multiple aspects, a calibration process integrating geometry, radiation, and spectrum is usually performed.

[0118] Specifically, the spectral environment is first determined based on the collected indoor spectral calibration data. Then, under the same spectral environment, indoor geometric calibration data is collected, a geometric calibration model is established, and geometric correction is performed. Finally, the collected indoor radiation calibration data is input into the geometric calibration model, and a radiation calibration model is established based on the corrected indoor radiation calibration data to perform radiation correction. Images captured by the multispectral camera are calibrated through these steps, and finally, an image that has undergone integrated spectral, geometric, and radiation calibration is output.

[0119] In one embodiment, see Figure 4 , the indoor and outdoor integrated calibration method also includes:

[0120] S401, collecting outdoor atmospheric calibration data, and inputting the outdoor atmospheric calibration data into a geometric correction model and a radiation correction model, and outputting the corrected outdoor atmospheric calibration data;

[0121] The outdoor atmospheric calibration data primarily includes aerial imagery of dark targets captured by drones and downlink radiance data collected by ground-based spectrometers. Both require a unified time system to record the acquisition time. Corrected outdoor atmospheric calibration data are geometrically distortion-free and allow for direct radiance reading.

[0122] S402, determining the atmospheric path radiation value according to the radiance value of the zero reflectivity target;

[0123] S403: Establish an atmospheric correction model based on the atmospheric radiation value and outdoor atmospheric calibration data. The atmospheric correction model is used to calibrate the image.

[0124] Specifically, based on the outdoor atmospheric calibration data taken from the outdoor aerial photography, the dark target plus empirical line method is adopted to establish fitting models for the radiance value of the zero reflectivity target and the ground downlink radiance value for each altitude. It should be noted that, in the open air, the radiance value of the zero reflectivity target can be considered as the atmospheric path radiation value. The radiance value of the zero reflectivity target can be obtained after correction by the geometric correction model and the radiation correction model. The radiance value corresponding to the central wavelength of the multispectral camera can be obtained from the radiance data synchronously measured by the ground object spectrometer, and then these radiance values ​​and GPS time (also known as GPS Time, GPST, global positioning time system, an atomic time reference composed of the GPS satellite-borne atomic clock and the ground monitoring station atomic clock) are interpolated with cubic splines. After interpolation, the ground downlink radiance value corresponding to each image collected by the UAV can be obtained. According to the atmospheric path radiation value and the downlink radiance value, the following fitting formula is set:

[0125]

[0126] Among them, L0 is the atmospheric radiation value, L p is the downlink radiance value, e is a natural constant, r and s are fitting coefficients.

[0127] By fitting the formula, an atmospheric correction model is established for image calibration.

[0128] In this embodiment, due to the large variations in atmospheric conditions in actual situations, remote sensing missions also need to collect atmospheric path radiation values ​​and downlink radiance values ​​during flight for atmospheric correction, thereby reducing the influence of atmospheric condition environmental data on the images captured by the multispectral camera and improving the accuracy of the calibration.

[0129] In one embodiment, see Figure 5 , the indoor and outdoor integrated calibration method also includes:

[0130] S501, collecting outdoor reflectivity calibration data, and inputting the outdoor reflectivity calibration data into the geometric correction model, the radiation correction model and the atmospheric correction model in sequence, and outputting the corrected outdoor reflectivity calibration data;

[0131] The outdoor reflectance calibration data collected by the equipment primarily includes downlink radiance data collected by a ground-based spectrometer and the imagery used for reflectance calibration, with a unified time system recording the acquisition time. Corrected outdoor reflectance calibration data is geometrically distortion-free and directly captures ground-truth radiance values.

[0132] S502. Obtain the downlink radiance value corresponding to the atmospheric path radiation value using the dark target method and the empirical linear method;

[0133] S503: Establish a reflectivity correction model based on the downlink radiance value and outdoor reflectivity calibration data. The reflectivity correction model is used to calibrate the image.

[0134] Specifically, the outdoor reflectivity calibration data image after geometric, radiometric, and atmospheric corrections is divided by the downlink radiance value at the corresponding time. The calculation formula is as follows:

[0135]

[0136] Among them, R(i) is the reflectivity of the target pixel, L(i) is the radiance of the target pixel after geometric and radiation correction, L0 is the atmospheric path radiation value, L p is the downlink radiance value.

[0137] Through the above calculation formula, a reflectivity correction model is established for image calibration.

[0138] In this embodiment, due to the large variations in atmospheric conditions in actual situations, remote sensing missions also need to collect atmospheric path radiation values ​​and downlink radiance values ​​during flight for emissivity correction, thereby reducing the influence of atmospheric condition environmental data on the images captured by the multispectral camera and improving the accuracy of calibration.

[0139] In one embodiment, the indoor and outdoor integrated calibration method further includes:

[0140] The calibration data of the image is input into the initial calibration model, and an image that meets the preset calibration data is output. The data type of the image calibration data is the same as that of the indoor calibration data.

[0141] In this embodiment, the images actually taken by the multispectral camera can be input into initial correction models such as geometry, radiation, atmosphere, and reflectivity according to actual needs to obtain images that meet the preset calibration data, thereby improving calibration efficiency and accuracy.

[0142] In one embodiment, step S102 includes:

[0143] Collect outdoor verification data at a portable outdoor calibration site;

[0144] Equipment in a movable outdoor calibration field should include at least one of a movable standard diffuse reflector target, a movable resolution target, a movable geometric target, and a movable zero-reflectivity target. These movable targets should be made of materials that are resistant to deformation over time and temperature, and the pattern resolution on the targets should be as high as possible. Furthermore, these targets should be lightweight, easy to transport, and have a bottom fixture.

[0145] Among them, the movable standard diffuse reflection targets can be selected from several types with reflectivity from 5% to 99%, such as standard diffuse reflection targets with reflectivity of 5%, 20%, 40%, 50%, 70%, and 99%; the movable resolution targets include Siemens star and three-line array targets; the movable geometric targets can select black and white cross targets, and the number of movable geometric targets is more than 30.

[0146] The movable zero-reflectivity target is a hollow cube painted with matte black paint. A hole is located at the top of the cube, and the interior is filled with black diffuse reflective cloth. Therefore, due to the blackbody effect, light entering the hole at the top of the dark target is almost completely absorbed, resulting in a reflectivity close to 0%.

[0147] In the movable outdoor calibration field, the movable standard diffuse reflector target, movable resolution target, and movable zero reflectivity target should be placed at the center of the movable outdoor calibration field. All movable geometric targets should be evenly fixed in a checkerboard pattern throughout the movable outdoor calibration field. The coordinates of the cross center position of the movable geometric targets should be measured using a total station and RTK equipment (real-time kinematic carrier phase differential technology equipment) to ensure that the movable geometric targets can be converted between the WGS 84 coordinate system (World Geodetic System - 1984 Coordinate System, an internationally adopted geocentric coordinate system) and the independently established coordinate system.

[0148] In addition, the collection of outdoor verification data and outdoor calibration data can be collected in the same movable outdoor calibration field.

[0149] In this embodiment, a scheme is proposed to collect outdoor verification data and calibration data in a movable outdoor calibration field, so that the multispectral camera can be calibrated smoothly while ensuring the maneuverability and flexibility of the low-altitude UAV.

[0150] In one embodiment, step S101 includes:

[0151] Collect indoor calibration data at the indoor calibration site;

[0152] The equipment in the indoor calibration field includes a monochromator, a standard radiation source and a geometric target;

[0153] Wherein, the standard radiation source includes a standard diffuse reflection plate and / or a standard integrating sphere;

[0154] The monochromator and the standard radiation source are placed on an optical platform and in a darkroom.

[0155] In this embodiment, indoor calibration data is collected in an indoor calibration field, and the indoor calibration data collection environment is fixed, so as to improve the accuracy of the collected data, thereby establishing a more accurate initial correction model and improving the calibration effect of the multispectral camera image.

[0156] In one embodiment, during the process of establishing the radiation correction model, two methods can be selected for vignetting effect calibration:

[0157] The first method is the lookup table method (LUT). This method takes the average value of the correction coefficient for each pixel position of all images. Similarly, a template with the same two-dimensional structure as the image is established based on the average value of each pixel. When correcting the vignetting effect, the pixel value of each position in the image to be corrected is multiplied by the corresponding correction coefficient in the template.

[0158] The second method is the binary cubic polynomial regression method, which fits the pixel position coordinates in the image and all correction coefficients using the following fitting formula:

[0159] VE=ax 3 +by 3 +cyx 2 +dxy 2 +ex 2 +fy 2 +gxy+hx+iy+j

[0160] Among them, VE is the correction coefficient output by the model, and a to j are fitting coefficients. When vignetting effect is calibrated, the correction coefficient is calculated according to the position coordinates in the image, and then the pixel value is multiplied by the correction coefficient.

[0161] In practice, if the correction accuracy needs to be given priority, the lookup table method is used to calibrate the vignetting effect; if the memory problem needs to be given priority, the binary cubic polynomial regression method can be used to calibrate the vignetting effect.

[0162] In this embodiment, the vignetting effect calibration is performed using a binary cubic polynomial regression method, which can reduce data memory pressure, improve the operating efficiency of the computing software, and further improve the efficiency of establishing the correction model.

[0163] In one embodiment, during the process of establishing the radiation correction model, two methods can be selected for quantum efficiency fitting:

[0164] The first method is the linear fitting method, which establishes a linear fitting relationship between the image pixel value and the radiance value. Each ISO or gain mode requires a corresponding fitting model. The fitting formula is as follows:

[0165] L(i0)=k1×DN(i0)+m1

[0166] Where L(i0) is the radiance value of the image principal point, DN(i0) is the pixel value of the image principal point, and k1 and m1 are fitting coefficients.

[0167] The second method is a nonlinear fitting method. Similarly, each ISO or gain mode requires a corresponding fitting model. The fitting formula is as follows:

[0168]

[0169] Where L(i0) is the radiance value of the image principal point, DN(i0) is the pixel value of the image principal point, and k, m, and n are fitting coefficients.

[0170] In practice, for photosensitive elements with good linearity in multispectral cameras, a linear fitting method is used to fit the quantum efficiency, while for photosensitive elements with poor linearity, a nonlinear fitting method can be used to fit the quantum efficiency.

[0171] In this embodiment, the quantum efficiency fitting is performed using a nonlinear fitting method, which can be applied to photosensitive elements with poor linearity in a multispectral camera, thereby using a more reasonable method to calibrate the components of the multispectral camera and further improve the calibration accuracy of the multispectral camera.

[0172] In one embodiment, the atmospheric correction model is established using the following formula:

[0173]

[0174] Among them, L0 is the atmospheric radiation value, L p is the downlink radiance value, e is a natural constant, r and s are fitting coefficients.

[0175] In this embodiment, the atmospheric correction model is established by the above formula, and the atmospheric correction is introduced into the calibration process of the low-altitude sensor to improve the observation accuracy in actual application, thereby improving the user experience of shooting images with a multispectral camera.

[0176] The following example uses the calibration of a narrow-band multispectral camera in the 675nm band to illustrate the specific implementation process of the indoor and outdoor integrated calibration method.

[0177] Figure 6 This is the overall flow chart of this case, which mainly includes establishing an indoor calibration field, establishing a movable outdoor calibration field, collecting data required for indoor geometry, radiation, and spectral calibration, collecting data required for outdoor geometry and radiation calibration, calculating geometry and spectral calibration parameters, establishing a geometry correction model, verifying the geometry correction model and making corrections, calculating radiation calibration parameters, establishing a radiation correction model, verifying the radiation correction model and making corrections, and establishing an atmosphere and reflectance correction model.

[0178] S601: Establish an indoor calibration field, mainly including: monochromator, standard radiation source, and geometric target.

[0179] S602: If Figure 7 As shown in the figure, a movable outdoor calibration field is established, which mainly includes several standard diffuse reflector targets, Siemens star targets, three-line array targets, several geometric targets, and zero reflectivity targets.

[0180] S603: Collect the data required for indoor geometry, radiometric, and spectral calibration. Specifically, a narrow-band multispectral camera with 16 ISO modes in the 675nm band and a fixed exposure time of 0.001 seconds is used. A checkerboard target is photographed from various distances and angles, with the target image covering approximately half of the field of view. A total of 20 images are captured to form the dataset required for indoor geometry calibration (i.e., indoor geometry calibration data).

[0181] In the absence of light and with the lens cap in place, five images were captured in each ISO mode, for a total of 80 images, forming the dataset required for indoor dark current calibration. The camera was set to ISO Auto mode, and the integrating sphere input current was set from 3.5A to 8A. An image was captured every 0.5A, for a total of ten images, forming the dataset required for indoor vignetting calibration. In each ISO mode, an image was captured every 0.1A of the integrating sphere input current, varying the camera's response from nearly underexposure to nearly overexposure. A total of 643 images were captured, forming the dataset required for indoor quantum efficiency fitting.

[0182] The monochromator was adjusted from 662 nm to 684 nm, and an image was taken every 0.1 nm. A total of 221 images were taken, which constituted the data set required for indoor spectral calibration (i.e., indoor spectral calibration data).

[0183] S604: Collect the data required for outdoor geometry and radiation calibration. Specifically: Use a ground-mounted camera to capture 20 images of a 99% standard diffuse reflector at 9:30, 11:00, and 14:30, with a 5-second interval between each capture. Simultaneously, use an ASD (handheld surface spectrometer) to collect the reflector radiance curve, collecting a total of 60 sets of data. Design an aerial data collection route, with the aircraft always facing north during collection, at altitudes of 90m, 200m, 250m, and 300m, with heading and lateral overlaps of 90% and 60%. Figure 8 This is an example image captured by a drone at a flight path of 90 meters. While simultaneously collecting images of the calibration field at four altitudes, an ASD handheld ground spectrometer was used on the ground to capture the radiance curve of a 99% standard reflectance diffuse reflectance panel every two seconds. A total of 128 images were captured of the outdoor calibration field. These ground-based and aerial images together constitute the dataset required for outdoor geometric and radiometric calibration.

[0184] S605: Calculate the geometric and spectral calibration parameters and establish a geometric calibration model. Specifically, based on the dataset required for indoor geometric calibration, use the checkerboard calibration method to perform geometric calibration of the multispectral camera and obtain the camera's image principal point coordinates, focal length, radial distortion coefficient, and eccentric distortion coefficient. Use the Gaussian fitting method to fit the pixel values ​​of the dataset required for indoor spectral calibration. The fitting results are as follows: Figure 9 As shown, the central wavelength is 674.25 nm, the central wavelength offset is within 1 nm, and the half-maximum width is 11.9 nm.

[0185] S606: Verify the geometric correction model and make corrections. Specifically, 5 images are extracted from each of the 4 flight altitudes in the data set required for outdoor geometric correction as a verification set (i.e., outdoor geometric correction data). The reprojection error is calculated to be 0.1 pixel (i.e., the preset error). It passes the test, so the result of the indoor geometric correction can be directly used for geometric correction (i.e., if the verification result is less than or equal to the preset error, the initial correction model is output, and the initial correction model is used to calibrate the image).

[0186] S607: Calculate radiation calibration parameters and establish a radiation calibration model. Specifically: Figure 10 This is a flowchart for using a lookup table (LUT) to correct dark current. Based on the data required for indoor dark current calibration and the lookup table method, dark current correction is performed on the image. In order to accurately correct the vignetting effect, a lookup table (LUT) is used in this case to correct the vignetting effect. Figure 11 The correction coefficient diagram, the X-axis and Y-axis represent the position of the pixel in the image. (X, Y) can determine the position coordinates of the pixel in the image, and the Z-axis represents the vignetting effect correction coefficient corresponding to the position coordinates. Figure 12a This is an example of an image before correction. Figure 12b is the corresponding corrected image example. Figure 12a and Figure 12b , the image before correction can be clearly seen to be brighter at the center edge and gradually darker from the center to the edge; but the brightness of the image after correction is flat and uniform. Since the linearity of the photosensitive element of the multispectral camera used in this case is poor, the quantum efficiency fitting adopts a nonlinear fitting method. Figure 13 This is the fitting result in ISO 100 mode.

[0187] S608: Verify the radiation correction model and make corrections. Specifically: 5 images are extracted from each of the three time periods in the data set required for outdoor ground radiation calibration as a verification set (i.e., outdoor radiation verification data). After the images in the verification set are corrected using the correction methods of S605 and S607, they are compared with the true radiance values ​​measured by the ground object spectrometer. The absolute percentage error is 8.32%, which passes the test. Therefore, the quantum efficiency fitting model in S607 can be used directly (i.e., if the verification result is less than or equal to the preset error, the initial correction model is output, and the initial correction model is used to calibrate the image).

[0188] S609: Establish an atmospheric and reflectivity correction model. Specifically, based on the dataset required for outdoor airborne radiation calibration, the image is corrected using the correction methods in S605 and S607 to obtain the radiance of a zero-reflectivity target, which can also be considered atmospheric path radiation. The radiance at 675 nm is interpolated from the radiance curve of the ground reflector collected at each flight altitude to obtain the ground downlink radiance at the time of image acquisition based on GPS time. An empirical line fitting method is used to fit the ground downlink radiance to the radiance of the zero-reflectivity target to obtain the atmospheric correction model. Figure 14 This is a schematic diagram of the atmospheric correction model for the 90m flight path. After accounting for atmospheric effects, the reflectance of the ground objects is obtained by dividing all pixels in the image by the ground downlink radiance at the corresponding time. Figure 15a This is an example of an image before calibration. Figure 15b This is an example of an image calibrated using the overall method in this case study. The pixel values ​​in the pre-calibrated image are the initial pixel values ​​after camera exposure, reflecting only the brightness of the object and lacking physical meaning. However, the pixel values ​​in the calibrated image are mapped to the physical quantity of radiance. The calibrated pixels are multiplied by 50,000, resulting in a 16-bit (pixel format byte) image.

[0189] Example 2

[0190] This embodiment provides an indoor and outdoor integrated calibration system, which is applied to the calibration of images taken by a multispectral camera and is used to implement the indoor and outdoor integrated calibration method of embodiment 1. Figure 16 , the indoor and outdoor integrated calibration system includes:

[0191] Acquisition module 1 is used to collect indoor calibration data, calculate indoor calibration data and establish an initial calibration model; wherein the initial calibration model includes a geometric calibration model and / or a radiation calibration model; and is also used to collect outdoor verification data, input the outdoor verification data into the initial calibration model, and obtain verification results; wherein the outdoor verification data and the indoor calibration data have the same data type;

[0192] Verification module 2 is used to output an initial correction model if the verification result is less than or equal to a preset error, and the initial correction model is used to calibrate the image.

[0193] In one embodiment, the acquisition module 1 is further used to acquire indoor spectral calibration data to determine the spectral environment; and is also used to acquire indoor geometric calibration data under the same spectral environment;

[0194] See also Figure 16 , the indoor and outdoor integrated calibration system also includes:

[0195] Calculation module 3 is used to calculate the corner points of indoor geometric calibration data according to the checkerboard calibration method, estimate the distortion coefficient using the least squares method, and then use the maximum likelihood method to optimize the estimation accuracy. The geometric correction model is established based on the corner points, distortion coefficient and estimation accuracy.

[0196] In one embodiment, the acquisition module 1 is further used to acquire indoor radiation calibration data, input the indoor radiation calibration data into the geometric correction model, and output the corrected indoor radiation calibration data;

[0197] The calculation module 3 is also used to perform dark current calibration, vignetting effect calibration and quantum efficiency fitting on the corrected indoor radiation calibration data, and establish a radiation correction model based on the processing results.

[0198] In one embodiment, see Figure 16 , the indoor and outdoor integrated calibration system also includes:

[0199] Correction module 4 is used to collect outdoor calibration data if the verification result is greater than the preset error, and correct the initial calibration model based on the indoor calibration data and the outdoor calibration data, where the data types of the outdoor calibration data and the indoor calibration data are the same.

[0200] In one embodiment, the acquisition module 1 is further used to acquire outdoor atmospheric calibration data, input the outdoor atmospheric calibration data into the geometric correction model and the radiation correction model, and output the corrected outdoor atmospheric calibration data;

[0201] The calculation module 3 is further used to determine the atmospheric path radiation value according to the radiance value of the zero reflectivity target; and is further used to establish an atmospheric correction model according to the atmospheric path radiation value and outdoor atmospheric calibration data. The atmospheric correction model is used to calibrate the image.

[0202] In one embodiment, the acquisition module 1 is further used to acquire outdoor reflectivity calibration data, and input the outdoor reflectivity calibration data into the geometric correction model, the radiation correction model and the atmospheric correction model in sequence, and output the corrected outdoor reflectivity calibration data;

[0203] Calculation module 3 is also used to obtain the downlink radiance value corresponding to the atmospheric path radiation value using the dark target method and the empirical linear method; a reflectance correction model is established based on the downlink radiance value and the outdoor reflectance calibration data, and the reflectance correction model is used to calibrate the image.

[0204] In one embodiment, see Figure 16 , the indoor and outdoor integrated calibration system also includes:

[0205] The calibration module 5 is used to input the calibration data of the image into the initial calibration model and output an image that meets the preset calibration data. The calibration data of the image is of the same data type as the indoor calibration data.

[0206] In one embodiment, the collection module 1 is further configured to collect outdoor verification data at a movable outdoor verification site;

[0207] The equipment of the movable outdoor calibration field includes at least one of a movable standard diffuse reflector target, a movable resolution target, a movable geometric target, and a movable zero reflectivity target;

[0208] The movable zero-reflectivity target is a hollow cube with a surface painted with black matte paint. A hole is set on the top of the cube, and the interior of the cube is filled with black diffuse reflection cloth.

[0209] In one embodiment, the acquisition module 1 is further used to collect indoor calibration data in an indoor calibration field;

[0210] The equipment in the indoor calibration field includes a monochromator, a standard radiation source and a geometric target;

[0211] Wherein, the standard radiation source includes a standard diffuse reflection plate and / or a standard integrating sphere;

[0212] The monochromator and the standard radiation source are placed on an optical platform and in a darkroom.

[0213] In one embodiment, the calculation module 3 is further configured to perform vignetting effect calibration using the following formula:

[0214] VE=ax 3 +by 3 +cyx 2 +dxy 2 +ex 2 +fy 2 +gxy+hx+iy+j

[0215] Among them, VE is the correction coefficient output by the model, a to j are fitting coefficients, and x and y are the position coordinates of all pixels in the image.

[0216] In one embodiment, the calculation module 3 is further configured to perform quantum efficiency fitting using the following formula:

[0217]

[0218] Where L(i0) is the radiance value of the image principal point, DN(i0) is the pixel value of the image principal point, and k, m, and n are fitting coefficients.

[0219] In one embodiment, the calculation module 3 is further configured to fit the atmospheric correction model using the following formula:

[0220]

[0221] Among them, L0 is the atmospheric radiation value, L p is the downlink radiance value, e is a natural constant, r and s are fitting coefficients.

[0222] It should be noted that the implementation method and technical effects of each module of the indoor and outdoor integrated calibration system of this embodiment can refer to the corresponding parts of Example 1 and will not be repeated here.

[0223] Example 3

[0224] This embodiment provides an electronic device, Figure 17 The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the indoor and outdoor integrated calibration method of Example 1 is implemented. Figure 17 The electronic device 30 shown is only an example and should not limit the functionality and scope of use of the embodiments of the present invention.

[0225] like Figure 17 As shown, the electronic device 30 may be a general-purpose computing device, such as a server device. Components of the electronic device 30 may include, but are not limited to, the at least one processor 31, the at least one memory 32, and a bus 33 connecting different system components (including the memory 32 and the processor 31).

[0226] The bus 33 includes a data bus, an address bus, and a control bus.

[0227] The memory 32 may include a volatile memory, such as a random access memory (RAM) 321 and / or a cache memory 322 , and may further include a read-only memory (ROM) 323 .

[0228] The memory 32 may also include a program / utility 325 having a set (at least one) of program modules 324, such program modules 324 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0229] The processor 31 executes various functional applications and data processing by running computer programs stored in the memory 32 , such as the indoor and outdoor integrated calibration method of embodiment 1 of the present invention.

[0230] The electronic device 30 may also communicate with one or more external devices 34 (e.g., a keyboard, a pointing device, etc.). Such communication may be performed via an input / output (I / O) interface 35. Furthermore, the model generating device 30 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 36. Figure 17 As shown, the network adapter 36 communicates with the other modules of the model-generated device 30 via the bus 33. It should be understood that, although not shown in the figures, other hardware and / or software modules may be used in conjunction with the model-generated device 30, including but not limited to microcode, device drivers, redundant processors, external disk drive arrays, RAID (RAID) systems, tape drives, and data backup storage systems.

[0231] It should be noted that although several units / modules or sub-units / modules of the electronic device are mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more units / modules described above may be embodied in a single unit / module. Conversely, the features and functions of a single unit / module described above may be further divided and embodied by multiple units / modules.

[0232] Example 4

[0233] This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the indoor and outdoor integrated calibration method of embodiment 1 is implemented.

[0234] The readable storage medium may include, but is not limited to, a portable disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0235] In a possible implementation manner, the present invention may also be implemented in the form of a program product, which includes program code. When the program product runs on a terminal device, the program code is used to enable the terminal device to execute the indoor and outdoor integrated calibration method of Example 1.

[0236] The program code for executing the present invention may be written in any combination of one or more programming languages, and may be executed entirely on the user device, partially on the user device, as a standalone software package, partially on the user device and partially on a remote device, or entirely on the remote device.

[0237] Although specific embodiments of the present invention have been described above, those skilled in the art will appreciate that these are merely illustrative and that the scope of the present invention is defined by the appended claims. Those skilled in the art may make various changes or modifications to these embodiments without departing from the principles and essence of the present invention, and such changes and modifications are intended to fall within the scope of the present invention.

Claims

1. An indoor and outdoor integrated calibration method, characterized in that: Applied to the calibration of images captured by a multispectral camera, the indoor and outdoor integrated calibration method includes: Collecting indoor calibration data, calculating the indoor calibration data and establishing an initial calibration model; wherein the initial calibration model includes a geometric calibration model and / or a radiation calibration model; Collecting outdoor verification data, inputting the outdoor verification data into the initial calibration model, and obtaining a verification result; wherein the data type of the outdoor verification data is the same as that of the indoor calibration data; If the verification result is less than or equal to the preset error, the initial correction model is output, and the initial correction model is used to calibrate the image; The steps of collecting indoor calibration data, calculating the indoor calibration data and establishing an initial calibration model include: Collecting indoor radiation calibration data, inputting the indoor radiation calibration data into the geometric correction model, and outputting corrected indoor radiation calibration data; Performing dark current calibration, vignetting effect calibration and quantum efficiency fitting on the corrected indoor radiation calibration data, and establishing a radiation correction model based on the processing results; The indoor and outdoor integrated calibration method further includes: Collecting outdoor atmospheric calibration data, inputting the outdoor atmospheric calibration data into the geometric correction model and the radiation correction model, and outputting corrected outdoor atmospheric calibration data; Determine the atmospheric path radiation value based on the radiance value of the zero reflectivity target; Establishing an atmospheric correction model based on the atmospheric radiation value and the outdoor atmospheric calibration data, wherein the atmospheric correction model is used to calibrate the image; The indoor and outdoor integrated calibration method further includes: Collecting outdoor reflectivity calibration data, and sequentially inputting the outdoor reflectivity calibration data into the geometric correction model, the radiation correction model, and the atmospheric correction model, and outputting corrected outdoor reflectivity calibration data; The dark target method and the empirical linear method are used to obtain the downlink radiance value corresponding to the atmospheric path radiation value; a reflectance correction model is established based on the downlink radiance value and the outdoor reflectance calibration data, and the reflectance correction model is used to calibrate the image.

2. The indoor and outdoor integrated calibration method according to claim 1, characterized in that: The steps of collecting indoor calibration data, calculating the indoor calibration data and establishing an initial calibration model include: Collect indoor spectrum calibration data to determine the spectrum environment; Under the same spectral environment, indoor geometric calibration data is collected, and the corner points of the indoor geometric calibration data are calculated according to the checkerboard calibration method. The distortion coefficient is estimated using the least squares method, and then the maximum likelihood method is used to optimize the estimation accuracy. A geometric correction model is established based on the corner points, the distortion coefficient and the estimation accuracy.

3. The indoor and outdoor integrated calibration method according to any one of claims 1 or 2, characterized in that: The steps of collecting outdoor verification data, inputting the outdoor verification data into the initial calibration model, and obtaining a verification result include: If the verification result is greater than the preset error, outdoor calibration data is collected, and the initial calibration model is corrected according to the indoor calibration data and the outdoor calibration data, where the data types of the outdoor calibration data and the indoor calibration data are the same.

4. The indoor and outdoor integrated calibration method according to claim 1, characterized in that: The indoor and outdoor integrated calibration method further includes: The calibration data of the image is input into the initial calibration model, and an image that meets the preset calibration data is output. The calibration data of the image has the same data type as the indoor calibration data.

5. The indoor and outdoor integrated calibration method according to claim 1, characterized in that: The step of collecting outdoor verification data includes: Collect outdoor verification data at a portable outdoor calibration site; The equipment of the movable outdoor calibration field includes at least one of a movable standard diffuse reflection plate target, a movable resolution target, a movable geometric target and a movable zero reflectivity target; The movable zero-reflectivity target is a hollow cube with a surface painted with black matte paint, a hole is set on the top of the cube, and the interior of the cube is filled with black diffuse reflection cloth.

6. The indoor and outdoor integrated calibration method according to claim 1, characterized in that: The step of collecting indoor calibration data includes: Collect indoor calibration data at the indoor calibration site; The equipment in the indoor calibration field includes a monochromator, a standard radiation source and a geometric target; Wherein, the standard radiation source includes a standard diffuse reflection plate and / or a standard integrating sphere; The monochromator and the standard radiation source are arranged on an optical platform and in a dark room.

7. The indoor and outdoor integrated calibration method according to claim 2, characterized in that: The indoor and outdoor integrated calibration method further includes: The following formula is used to perform the vignetting effect correction: Among them, VE is the correction coefficient output by the model, a to j are fitting coefficients, and x and y are the position coordinates of all pixels in the image.

8. The indoor and outdoor integrated calibration method according to claim 2, characterized in that: The indoor and outdoor integrated calibration method further includes: The quantum efficiency fitting is performed using the following formula: in, is the radiance value of the image principal point, is the pixel value of the principal point, 、 and is the fitting coefficient.

9. The indoor and outdoor integrated calibration method according to claim 2, characterized in that: The indoor and outdoor integrated calibration method further includes: The atmospheric correction model is fitted using the following formula: in, is the atmospheric radiation value, is the downlink radiance value, is a natural constant, r and s are fitting coefficients.

10. An indoor and outdoor integrated calibration system, characterized in that: Applied to the calibration of images captured by a multispectral camera, the indoor and outdoor integrated calibration system implements the indoor and outdoor integrated calibration method according to any one of claims 1 to 9, and the indoor and outdoor integrated calibration system includes: an acquisition module, configured to acquire indoor calibration data, calculate the indoor calibration data, and establish an initial calibration model; wherein the initial calibration model includes a geometric calibration model and / or a radiation calibration model; and further configured to acquire outdoor verification data, input the outdoor verification data into the initial calibration model, and obtain a verification result; wherein the outdoor verification data is of the same data type as the indoor calibration data; a verification module, configured to output the initial correction model if the verification result is less than or equal to a preset error, wherein the initial correction model is used to calibrate the image; The acquisition module is further used to acquire indoor radiation calibration data, input the indoor radiation calibration data into the geometric correction model, and output the corrected indoor radiation calibration data; The calculation module is also used to perform dark current calibration, vignetting effect calibration and quantum efficiency fitting on the corrected indoor radiation calibration data, and establish a radiation correction model based on the processing results; The acquisition module is further used to acquire outdoor atmospheric calibration data, input the outdoor atmospheric calibration data into the geometric correction model and the radiation correction model, and output the corrected outdoor atmospheric calibration data; The calculation module is further used to determine the atmospheric path radiation value based on the radiance value of the zero reflectivity target; and is further used to establish an atmospheric correction model based on the atmospheric path radiation value and outdoor atmospheric calibration data, and the atmospheric correction model is used to calibrate the image; The acquisition module is further used to acquire outdoor reflectivity calibration data, and input the outdoor reflectivity calibration data into the geometric correction model, the radiation correction model and the atmospheric correction model in sequence, and output the corrected outdoor reflectivity calibration data; The calculation module is also used to obtain the downlink radiance value corresponding to the atmospheric path radiation value using the dark target method and the empirical linear method; and to establish a reflectance correction model based on the downlink radiance value and outdoor reflectance calibration data, and the reflectance correction model is used to calibrate the image.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the indoor and outdoor integrated calibration method according to any one of claims 1 to 9 is implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the indoor and outdoor integrated calibration method according to any one of claims 1 to 9 is implemented.

Images