A method for stereo image stitching based on UAV hyperspectral imager

Abstract

This invention relates to the field of hyperspectral imaging technology, specifically to a method for stitching stereoscopic images based on a UAV hyperspectral imager. The method includes the following steps: acquiring state parameters of image data collected by the UAV hyperspectral imager, determining the state during imaging to obtain state parameter signals, and then calculating imaging state values ​​based on the state parameters; acquiring region parameters of image data collected by the UAV hyperspectral imager, determining the region during imaging to obtain region parameter signals, and then calculating imaging region values ​​based on the region parameters; performing fusion analysis based on the imaging state values ​​and imaging region values ​​to obtain image stitching selection values; acquiring imaging quality values, further filtering the imaging data to obtain stitched image data; this invention effectively improves the visual consistency and coherence of the stitched image by selecting the image data to be stitched.

Description

Technical Field

[0001] This invention relates to the field of hyperspectral imaging technology, and more specifically to a method for stitching stereoscopic images based on a UAV hyperspectral imager. Background Technology

[0002] The stereo image stitching method based on UAV hyperspectral imagers involves several key steps that ensure high-quality image stitching, thereby providing accurate and detailed ground feature information. The detailed steps of this method are as follows:

[0003] I. Data Collection

[0004] First, drones equipped with hyperspectral imagers are used for flight operations to collect hyperspectral image data of the target area. During flight, the drone needs to maintain a stable altitude and speed to ensure the consistency and comparability of the collected image data.

[0005] II. Pretreatment

[0006] The collected raw hyperspectral image data needs to be preprocessed to remove noise, correct distortion, and improve image quality. Preprocessing steps include radiometric calibration, atmospheric correction, and geometric correction. These steps help eliminate the influence of environmental factors on the image data and improve the accuracy of subsequent processing.

[0007] III. Stereoscopic Matching

[0008] Next, stereo matching is performed, a crucial step in image stitching. By comparing image data collected from different perspectives, corresponding points of the same ground features are found, thus establishing stereo correspondences. This step requires the use of computer vision and image processing techniques to achieve accurate matching and the establishment of correspondences.

[0009] IV. Splicing and Fusion

[0010] Based on stereo matching, image data from different perspectives is stitched together. Using a specific algorithm, image data is merged to form a complete stereo map. During the stitching process, factors such as color, brightness, and contrast of the images need to be considered to ensure visual consistency and coherence in the stitched image.

[0011] Therefore, we can see that image data collection is the source and an important part of the stereoscopic image stitching process, and the quality of image data collection will affect the quality of stereoscopic image stitching.

[0012] Chinese patent CN110111255B discloses a stereo image stitching method, which includes the following steps: inputting left and right stereo images, pre-aligning the left and right images by minimizing the disparity homography and reducing the vertical disparity; further aligning the left and right images by multi-constraint deformation, while introducing an initial disparity map to maintain the disparity consistency of the stereo images; determining the optimal seam based on the seam selection and fusion of disparity consistency and performing stereo image stitching.

[0013] In existing technologies, UAV hyperspectral imagers cannot progressively filter the acquired image data based on factors such as the state and quality of the image during acquisition. This is not conducive to subsequent stereoscopic image stitching and cannot effectively improve the quality of stereoscopic image stitching. Summary of the Invention

[0014] The purpose of this invention is to provide a stereo image stitching method based on a UAV hyperspectral imager. The technical problem solved by this invention is that the obtained image data cannot be progressively screened according to factors such as the state and quality of the image during imaging, which is not conducive to subsequent stereo image stitching and cannot effectively improve the quality of stereo image stitching.

[0015] The objective of this invention can be achieved through the following technical solutions:

[0016] A method for stitching stereo images based on a UAV hyperspectral imager includes: data collection and stitching fusion. The data collection process includes the following steps:

[0017] Step 1: Obtain the state parameters of the image data acquired by the UAV hyperspectral imager, determine the state during imaging to obtain the state parameter signal, and then calculate the imaging state value based on the state parameters;

[0018] The state parameters include hovering position and imaging angle;

[0019] Step 2: Obtain the regional parameters of the image data collected by the UAV hyperspectral imager, determine the region during imaging to obtain the regional parameter signal, and then calculate the imaging region value based on the regional parameters;

[0020] Among them, the region parameters include the imaging region range;

[0021] Step 3: Perform fusion analysis based on imaging state values ​​and imaging region values ​​to obtain the screening values ​​for image stitching;

[0022] Step 4: Based on the image stitching screening value ZSq obtained at each acquisition point of the imaging trajectory, obtain the imaging quality value, and further screen the imaging data to obtain the stitched image data.

[0023] As a further aspect of the present invention: in step 1, the process of acquiring the state parameter signal is as follows:

[0024] If the hovering position is within the hovering position range, a hovering pass signal is generated; if the imaging angle is within the imaging angle range, an angle pass signal is generated.

[0025] If both hovering and angle pass signals are obtained simultaneously, a high-quality imaging status signal is generated.

[0026] As a further aspect of the present invention: the calculation process of the imaging state value is as follows:

[0027] When a high-quality imaging state signal is obtained, the hover position deviation ratio and imaging angle deviation ratio are acquired. The hover position deviation ratio and imaging angle deviation ratio are summed to obtain the imaging state value.

[0028] Specifically, the process of obtaining the hover position deviation ratio is as follows:

[0029] Obtain the distance between the hover position and the center point of the hover position range, mark it as the hover position deviation value, divide the hover position deviation value by the hover position range, and calculate the hover position deviation ratio;

[0030] The process of obtaining the imaging angle deviation ratio is as follows:

[0031] The difference between the obtained imaging angle and the median value of the imaging angle range is calculated to obtain the imaging angle offset value. The imaging angle offset value is then divided by the imaging angle range value to calculate the imaging angle deviation ratio.

[0032] As a further aspect of the present invention: in step 2, the process of acquiring the regional parameter signal is as follows:

[0033] The current imaging region is compared with the previous imaging region to obtain the area of ​​the overlapping region.

[0034] If the area of ​​the overlapping region is greater than or equal to the threshold of the overlapping region area, a qualified imaging region signal is generated.

[0035] As a further aspect of the present invention: the process of obtaining the imaging region value is as follows:

[0036] When a qualified signal of the imaging region is obtained, the imaging region area ratio and imaging region alignment ratio are acquired. The imaging region value is obtained by subtracting the imaging region area ratio from the imaging region alignment ratio.

[0037] The process of obtaining the imaging region area ratio is as follows:

[0038] Divide the area of ​​the overlapping region by the threshold of the overlapping region to obtain the area ratio of the imaging region;

[0039] The process of obtaining the imaging region homogeneity ratio is as follows:

[0040] Obtain the endpoints of the overlapping region, connect the endpoints end to end to obtain the imaging region line length, then obtain the actual boundary line length of the overlapping region, divide the actual boundary line length of the overlapping region by the imaging region line length to obtain the imaging region alignment ratio.

[0041] As a further aspect of the present invention: In step 3, the imaging state value and the current imaging area value are obtained and marked as ZZ and ZQ respectively. The imaging stitching screening value ZSq is calculated by the formula ZSq=a1*ZZ+a2*ZQ, where a1 and a2 are both proportional coefficients.

[0042] The image data acquired by the hyperspectral imager is filtered by the imaging stitching screening value ZSq, and image data with a larger imaging stitching screening value ZSq is selected first.

[0043] As a further aspect of the present invention: in step 4, all image data at each acquisition point on the imaging trajectory are acquired, arranged from largest to smallest according to the imaging stitching screening value ZSq, and a preset initial screening quantity is set, and the image data at the end of the arrangement is deleted to obtain the initial screening image;

[0044] Based on the initial image screening, the image brightness ratio and image contrast ratio are obtained. The image brightness ratio and image contrast ratio are added together to obtain the image quality value.

[0045] The initial screening images from each acquisition point are arranged in descending order of imaging quality value, and a preset initial screening quantity is set. The image data at the bottom of the list is deleted to obtain the images for further screening.

[0046] As a further aspect of the present invention: a three-dimensional coordinate system is constructed with the acquisition point location as the X-axis, the image stitching screening value as the Y-axis, and the image quality value as the Z-axis. The acquisition point location, image stitching screening value, and image quality value corresponding to the image to be screened again are substituted into the three-dimensional coordinate system to obtain the image coordinate point.

[0047] By extracting formulas Image data from each acquisition point location in the three-dimensional coordinate system that satisfies the extraction formula is extracted to obtain stitched image data; where, in the extraction formula, L i-1 D represents the distance difference between imaging coordinate points at adjacent acquisition points. i This represents the distance from the imaging coordinate point at the acquisition point location to the acquisition point on the X-axis, i.e. y i The Z represents the y-axis coordinate value of the imaging coordinate point. iThis represents the z-axis coordinate value of the imaging point, i represents the number of acquisition points, and LY represents all L... i-1 The threshold for summation, DY represents all D i The threshold for summation.

[0048] As a further aspect of the present invention: the process of obtaining the image brightness ratio is as follows:

[0049] Divide the obtained image brightness by the image brightness threshold to obtain the image brightness ratio.

[0050] As a further aspect of the present invention: the process of obtaining the image contrast ratio is as follows:

[0051] Divide the obtained image contrast by the image contrast threshold to obtain the image contrast ratio.

[0052] The beneficial effects of this invention are:

[0053] (1) In the process of monitoring the collected hyperspectral image data, this invention obtains the state parameters of the image data collected by the UAV hyperspectral imager, judges the state during imaging to obtain the state parameter signal, and then calculates the imaging state anomaly value based on the state parameter; obtains the region parameters of the image data collected by the UAV hyperspectral imager, judges the region during imaging to obtain the region parameter signal, and then calculates the imaging region anomaly value based on the region parameter; and performs fusion analysis based on the imaging state value and the imaging region value to obtain the imaging stitching screening value; the image data collected by the hyperspectral imager can be screened, so that the screened image data is conducive to subsequent stereoscopic image stitching and improves the quality of stereoscopic image stitching.

[0054] (2) Based on the imaging quality value ZSq obtained at each acquisition point of the imaging trajectory, the present invention obtains the imaging quality value and performs further screening of the imaging data. The present invention further filters the image data to be stitched by combining the imaging quality value and the imaging quality value in sequence. Then, by combining the imaging quality value and the imaging quality value in three dimensions, the image data to be stitched is further filtered, thereby effectively improving the visual consistency and coherence of the stitched image. Attached Figure Description

[0055] The invention will now be further described with reference to the accompanying drawings.

[0056] Figure 1 This is a flowchart of the stereoscopic image splicing method in Embodiment 1 of the present invention;

[0057] Figure 2 This is a flowchart of the real-time monitoring process in Embodiment 1 of the present invention;

[0058] Figure 3 This is a flowchart of the real-time monitoring process in Embodiment 2 of the present invention. Detailed Implementation

[0059] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0060] Example 1

[0061] Please see Figure 1 As shown, this invention is a method for stitching stereoscopic images based on a UAV hyperspectral imager, comprising the following steps:

[0062] Step 1: Data Collection

[0063] First, drones equipped with hyperspectral imagers are used for flight operations to collect hyperspectral image data of the target area; the collected hyperspectral image data is then monitored in real time.

[0064] Step 2: Splicing and Fusion

[0065] By comparing image data collected from different perspectives, corresponding points of the same ground features are found, thereby establishing a three-dimensional correspondence; image data from different perspectives are then fused together to form a complete three-dimensional map.

[0066] It should be noted that the splicing and fusion method can adopt a stereoscopic image splicing method as disclosed in the patent document with publication number CN110111255B.

[0067] Please see Figure 2 As shown, the real-time monitoring process in step one includes the following steps:

[0068] Step 1: Obtain the state parameters of the image data acquired by the UAV hyperspectral imager, determine the state during imaging to obtain the state parameter signal, and then calculate the imaging state value based on the state parameters;

[0069] The state parameters include hovering position and imaging angle;

[0070] In some embodiments, when the UAV hyperspectral imager moves along the imaging trajectory, the hovering position and imaging angle of the UAV hyperspectral imager are acquired;

[0071] The hover position is compared with the hover position range. If the hover position is within the hover position range, a hover pass signal is generated. If the hover position is not within the hover position range, a hover fail signal is generated.

[0072] The imaging angle is compared with the imaging angle range. If the imaging angle is within the imaging angle range, an angle qualified signal is generated. If the imaging angle is not within the imaging angle range, an angle unqualified signal is generated.

[0073] The hovering qualified signal, hovering unqualified signal, angle qualified signal and angle unqualified signal are combined and analyzed. That is, if both hovering qualified signal and angle qualified signal are obtained at the same time, a high-quality imaging state signal is generated; otherwise (otherwise means that one or both of hovering qualified signal and angle qualified signal are included), a low-quality imaging state signal is generated.

[0074] When a high-quality imaging state signal is obtained, the hover position deviation ratio and imaging angle deviation ratio are acquired. The hover position deviation ratio and imaging angle deviation ratio are summed to obtain the imaging state value.

[0075] Specifically, the process of obtaining the hover position deviation ratio is as follows:

[0076] Obtain the distance between the hover position and the center point of the hover position range, mark it as the hover position deviation value, divide the hover position deviation value by the hover position range, and calculate the hover position deviation ratio;

[0077] The process of obtaining the imaging angle deviation ratio is as follows:

[0078] The difference between the obtained imaging angle and the median value of the imaging angle range is calculated to obtain the imaging angle offset value. The imaging angle offset value is then divided by the imaging angle range value to calculate the imaging angle deviation ratio.

[0079] It should be explained that the degree of deviation between the hovering position and the imaging angle during hyperspectral imaging can effectively determine the stability of the hyperspectral imager during imaging acquisition. That is, the smaller the hovering position deviation ratio and the imaging angle deviation ratio, the higher the quality of the image data acquired by the hyperspectral imager.

[0080] Step 2: Obtain the regional parameters of the image data collected by the UAV hyperspectral imager, determine the region during imaging to obtain the regional parameter signal, and then calculate the imaging region value based on the regional parameters;

[0081] Among them, the region parameters include the imaging region range;

[0082] In some embodiments, the imaging area range of the UAV hyperspectral imager is acquired when the UAV hyperspectral imager moves along the imaging trajectory;

[0083] The current imaging region is compared with the previous imaging region to obtain the area of ​​the overlapping region.

[0084] Compare the area of ​​the overlapping region with the threshold for the area of ​​the overlapping region;

[0085] If the area of ​​the overlapping region is greater than or equal to the threshold of the overlapping region, a qualified imaging region signal is generated.

[0086] If the area of ​​the overlapping region is smaller than the threshold for the area of ​​the overlapping region, an unqualified imaging region signal is generated.

[0087] When a qualified signal of the imaging region is obtained, the imaging region area ratio and imaging region alignment ratio are acquired. The imaging region value is obtained by subtracting the imaging region area ratio from the imaging region alignment ratio.

[0088] Specifically, the process of obtaining the imaging region area ratio is as follows:

[0089] Divide the area of ​​the overlapping region by the threshold of the overlapping region to obtain the area ratio of the imaging region.

[0090] The process of obtaining the imaging region homogeneity ratio is as follows:

[0091] Obtain the endpoints of the overlapping region, connect the endpoints end to end to obtain the imaging region line length, then obtain the actual boundary line length of the overlapping region, divide the actual boundary line length of the overlapping region by the imaging region line length to obtain the imaging region alignment ratio.

[0092] It needs to be explained that when imaging with a hyperspectral imager, the imaging area is obtained from adjacent acquisition points. By using the imaging area ratio and imaging area alignment ratio, the continuity of adjacent images during hyperspectral imager acquisition can be effectively judged. That is, the larger the overlapping area between adjacent images, the more the actual boundary of the overlapping area tends to be a straight line, indicating that the adjacent images have good continuity, which is convenient for subsequent image stitching and overlapping.

[0093] Step 3: Perform fusion analysis based on imaging state values ​​and imaging region values ​​to obtain the screening values ​​for image stitching;

[0094] In some embodiments, the imaging state value and the current imaging region value are obtained and labeled as ZZ and ZQ, respectively. The imaging stitching screening value ZSq is calculated by the formula ZSq=a1*ZZ+a2*ZQ, where a1 and a2 are both proportional coefficients, a1+a2=1, a1 takes the value of 0.62, and a2 takes the value of 0.38. The values ​​of a1 and a2 represent the relative importance of the current imaging state value and the current imaging region value on the current imaging stitching screening value ZSq.

[0095] It can filter the image data acquired by the hyperspectral imager by using the imaging stitching screening value ZSq, and prioritize the image data with a larger imaging stitching screening value ZSq. This image data is beneficial for subsequent stereo image stitching and improves the quality of stereo image stitching.

[0096] It should be noted that fusing imaging state values ​​and imaging region values ​​facilitates the filtering of image data;

[0097] The technical solution of this invention is as follows: During the monitoring of collected hyperspectral image data, the state parameters of the image data acquired by the UAV hyperspectral imager are obtained, and the state parameters are obtained by judging the state during imaging. Then, anomaly values ​​of the imaging state are calculated based on the state parameters. The region parameters of the image data acquired by the UAV hyperspectral imager are obtained, and the region parameters are obtained by judging the region during imaging. Then, anomaly values ​​of the imaging region are calculated based on the region parameters. The imaging state values ​​and imaging region values ​​are fused and analyzed to obtain the imaging stitching screening values. The image data acquired by the hyperspectral imager can be screened, so that the screened image data is conducive to subsequent stereoscopic image stitching and improves the quality of stereoscopic image stitching.

[0098] Example 2

[0099] Please see Figure 3 As shown, the real-time monitoring process in step one also includes the following steps:

[0100] Step 4: Based on the image stitching screening value ZSq obtained at each acquisition point of the imaging trajectory, obtain the imaging quality value, and further screen the imaging data to obtain the stitched image data;

[0101] In some embodiments, all image data at each acquisition point on the imaging trajectory are acquired, arranged from largest to smallest according to the imaging stitching screening value ZSq, and a preset initial screening quantity is set. Image data at the end of the arrangement is deleted to obtain a preliminary screened image. Preferably, the preset initial screening quantity is 1 / 2 of all image data at each acquisition point.

[0102] Based on the initial image screening, the image brightness ratio and image contrast ratio are obtained. The image brightness ratio and image contrast ratio are added together to obtain the image quality value.

[0103] The initial filtered images from each acquisition point are arranged in descending order of imaging quality value, and a preset initial filtering quantity is set. The image data at the end of the ranking is deleted to obtain images for further filtering. Preferably, the preset second filtering quantity is 1 / 2 of all image data at each acquisition point.

[0104] A three-dimensional coordinate system is constructed with the acquisition point location as the X-axis, the image stitching selection value as the Y-axis, and the image quality value as the Z-axis. The acquisition point location, image stitching selection value, and image quality value corresponding to the image to be further selected are substituted into the three-dimensional coordinate system to obtain the image coordinate points.

[0105] By extracting formulas Image data from each acquisition point location in the three-dimensional coordinate system that satisfies the extraction formula is extracted to obtain stitched image data; where, in the extraction formula, L i-1 D represents the distance difference between imaging coordinate points at adjacent acquisition points. i This represents the distance from the imaging coordinate point at the acquisition point location to the acquisition point on the X-axis, i.e. y i The Z represents the y-axis coordinate value of the imaging coordinate point. i This represents the z-axis coordinate value of the imaging point, i represents the number of acquisition points, and LY represents all L... i-1 The threshold for summation, DY represents all D i Threshold for summation;

[0106] The obtained stitched image data is sent to the stitching and fusion step to complete the stitching of the image data;

[0107] It should be noted that the significance of the extraction formula is as follows: among many re-selected images, the formula with values ​​less than or equal to LY is used to extract images with small differences in image data performance between adjacent acquisition locations, and the formula with values ​​greater than or equal to DY is used to extract images with good image data performance among all re-selected images.

[0108] The extracted image data coordinates can be used for stitching images, which can effectively improve the visual consistency and coherence of the stitched image.

[0109] Specifically, the process of obtaining the image brightness ratio is as follows:

[0110] Divide the obtained image brightness by the image brightness threshold to obtain the image brightness ratio;

[0111] The process of obtaining the image contrast ratio is as follows:

[0112] Divide the obtained image contrast by the image contrast threshold to get the image contrast ratio. Image contrast refers to the measurement of different brightness levels between the brightest white and the darkest black in the bright and dark areas of an image, reflecting the magnitude of the grayscale contrast of the image.

[0113] The technical solution of this invention is as follows: Based on the imaging stitching screening value ZSq obtained at each acquisition point of the imaging trajectory, an imaging quality value is obtained, and the imaging data is further screened. This invention further screens the image data to be stitched by combining the imaging stitching screening value and the imaging quality value in a sequential manner. Then, by combining the imaging stitching screening value and the imaging quality value in three dimensions, the image data to be stitched is further screened, thereby effectively improving the visual consistency and coherence of the stitched image.

[0114] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0115] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for stitching stereoscopic images based on a UAV hyperspectral imager, comprising the following steps: Data collection, splicing, and fusion, characterized by the following steps during the data collection process: Step 1: Obtain the state parameters of the image data acquired by the UAV hyperspectral imager, determine the state during imaging to obtain the state parameter signal, and then calculate the imaging state value based on the state parameters; The state parameters include hovering position and imaging angle; Step 2: Obtain the regional parameters of the image data collected by the UAV hyperspectral imager, determine the region during imaging to obtain the regional parameter signal, and then calculate the imaging region value based on the regional parameters; Among them, the region parameters include the imaging region range; Step 3: Perform fusion analysis based on imaging state values ​​and imaging region values ​​to obtain the screening values ​​for image stitching; Step 4: Based on the image stitching screening values ​​obtained at each acquisition point of the imaging trajectory, obtain the imaging quality value, and further screen the imaging data to obtain the stitched image data; In step 1, the process of acquiring the state parameter signal is as follows: If the hovering position is within the hovering position range, a hovering pass signal is generated; if the imaging angle is within the imaging angle range, an angle pass signal is generated. If both hovering and angle pass signals are obtained simultaneously, a high-quality imaging status signal is generated. The calculation process for the imaging state value is as follows: When a high-quality imaging state signal is obtained, the hover position deviation ratio and imaging angle deviation ratio are acquired. The hover position deviation ratio and imaging angle deviation ratio are summed to obtain the imaging state value. Specifically, the process of obtaining the hover position deviation ratio is as follows: Obtain the distance between the hover position and the center point of the hover position range, mark it as the hover position deviation value, divide the hover position deviation value by the hover position range, and calculate the hover position deviation ratio; The process of obtaining the imaging angle deviation ratio is as follows: The difference between the obtained imaging angle and the median value of the imaging angle range is calculated to obtain the imaging angle offset value. The imaging angle offset value is then divided by the imaging angle range value to calculate the imaging angle deviation ratio. In step 4, all image data at each acquisition point on the imaging trajectory are acquired, arranged from largest to smallest according to the imaging stitching screening value ZSq, and a preset initial screening quantity is set. Image data at the end of the arrangement is deleted to obtain the initial screened image. Based on the initial image screening, the image brightness ratio and image contrast ratio are obtained. The image brightness ratio and image contrast ratio are added together to obtain the image quality value. The initial screening images from each acquisition point are arranged in descending order of imaging quality value, and a preset number of initial screenings is set. The image data at the end of the list is deleted to obtain images for further screening. A three-dimensional coordinate system is constructed with the acquisition point location as the X-axis, the image stitching selection value as the Y-axis, and the image quality value as the Z-axis. The acquisition point location, image stitching selection value, and image quality value corresponding to the image to be further selected are substituted into the three-dimensional coordinate system to obtain the image coordinate points. By extracting formulas Extract image data from each acquisition point in the three-dimensional coordinate system that satisfies the extraction formula to obtain stitched image data; In the extraction formula, L i-1 D represents the distance difference between imaging coordinate points at adjacent acquisition points. i This represents the distance from the imaging coordinate point at the acquisition point location to the acquisition point on the X-axis, i.e. ,y i The z-axis coordinates of the imaging points represent the y-axis coordinates. i This represents the z-axis coordinate value of the imaging point, i represents the number of acquisition points, and LY represents all L... i-1 The threshold for summation, DY represents all D i The threshold for summation.

2. The method for stitching stereoscopic images based on a UAV hyperspectral imager according to claim 1, characterized in that, In step 2, the process of acquiring the regional parameter signal is as follows: The current imaging region is compared with the previous imaging region to obtain the area of ​​the overlapping region. If the area of ​​the overlapping region is greater than or equal to the threshold of the overlapping region area, a qualified imaging region signal is generated.

3. The method for stitching stereoscopic images based on a UAV hyperspectral imager according to claim 2, characterized in that, The process of obtaining the imaging region value is as follows: When a qualified signal of the imaging region is obtained, the imaging region area ratio and imaging region alignment ratio are acquired. The imaging region value is obtained by subtracting the imaging region area ratio from the imaging region alignment ratio. The process of obtaining the imaging region area ratio is as follows: Divide the area of ​​the overlapping region by the threshold of the overlapping region to obtain the area ratio of the imaging region; The process of obtaining the imaging region homogeneity ratio is as follows: Obtain the endpoints of the overlapping region, connect the endpoints end to end to obtain the imaging region line length, then obtain the actual boundary line length of the overlapping region, divide the actual boundary line length of the overlapping region by the imaging region line length to obtain the imaging region alignment ratio.

4. The method for stitching stereoscopic images based on a UAV hyperspectral imager according to claim 1, characterized in that, In step 3, the imaging state value and the current imaging area value are obtained and labeled as ZZ and ZQ respectively, and then processed using the formula... The image stitching selection value ZSq is calculated, where a1 and a2 are both proportional coefficients; The image data acquired by the hyperspectral imager is filtered by the imaging stitching screening value ZSq, and image data with a larger imaging stitching screening value ZSq is selected first.

5. The method for stitching stereoscopic images based on a UAV hyperspectral imager according to claim 1, characterized in that, The process of obtaining the image brightness ratio is as follows: Divide the obtained image brightness by the image brightness threshold to obtain the image brightness ratio.

6. The method for stitching stereoscopic images based on a UAV hyperspectral imager according to claim 1, characterized in that, The process of obtaining the image contrast ratio is as follows: Divide the obtained image contrast by the image contrast threshold to obtain the image contrast ratio.

Images