Remote field investigation data informatization processing method and system

Abstract

The invention discloses a remote field investigation data informatization processing method and system. The method comprises the following steps: collecting a plurality of pictures at different heights and different shooting angles of a target position; splicing the plurality of acquired pictures into a panoramic image through a panoramic image synthesis algorithm; and uploading the synthesized panoramic image to a pre-constructed field investigation data management platform to realize the viewing, management and editing functions of the panoramic image, and carrying out association link on the panoramic image and the electronic document data of the corresponding project. According to the invention, the panoramic picture synthesis technology is utilized to apply the panoramic picture to the field investigation of the railway engineering, more visual and vivid image data is provided for the field scene, the problem of scattered data of the field data of the railway engineering is solved, and the efficiency of sorting and later viewing of the field data is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of surveying, and particularly to a method and system for information processing of remote field survey data. Background Art

[0002] In recent years, the informatization development of the survey and design industry has gone through multiple stages from network construction to the application of "Internet +" in survey and design. Some survey and design enterprises use new generation information technologies such as "cloud, big data, Internet of Things, artificial intelligence, and mobile Internet" (cloud computing, big data, Internet of Things, artificial intelligence, mobile Internet) to improve the survey quality and accelerate the pace of informatization. Previous researchers have focused on the control of the quality of field surveys, proposed solutions to strengthen on-site control using mobile broadband technology, and summarized and analyzed the main characteristics of survey informatization technologies, all of which have made progress in the research of key survey informatization technologies and provided useful references for the construction of survey informatization.

[0003] Currently, the informatization management of field surveys for railway projects is still in its infancy, with disadvantages such as simple functions, complex operations, scattered modules, a bias towards supervision, and poor promotion and usability. As a result, the complex field survey data lacks relevance. At the same time, current field survey image data mainly consists of two-dimensional photos, making it impossible to intuitively and vividly understand the surrounding environment of the survey location. Therefore, it is necessary to deeply analyze the pain points of field work, sort out the relationship between field and indoor surveys, and on this basis, scientifically build the framework of an informatization management system for engineering surveys to provide a relatively reliable solution for the whole process informatization of engineering surveys. Summary of the Invention

[0004] This application provides a method and system for information processing of remote field survey data to solve the disadvantages of the existing field survey data for railway projects being complex, lacking relevance, and not being intuitive and vivid enough, and further solve the problems of the single form of field survey image data and the unified integration and management of field survey data.

[0005] According to a first aspect, in one embodiment, a method for information processing of remote field survey data is provided. The method includes:

[0006] Collect multiple pictures at different heights and different shooting angles of a target location;

[0007] Stitch the collected multiple pictures into a panoramic image through a panoramic image synthesis algorithm;

[0008] Upload the synthesized panoramic image to a pre-constructed data management platform for field survey data, to implement functions such as viewing, management, and editing of the panoramic image, and to associate and link the panoramic image with the corresponding project electronic document data.

[0009] Further, collecting multiple pictures at different heights and different shooting angles of a target location specifically includes:

[0010] Fly the drone carrying the camera to the designated target position and determine the hovering height according to the height of the on-site building;

[0011] At different hovering heights, adjust the gimbal angle, and collect multiple photos by circling 360 degrees respectively at different preset angles of the camera horizontally and obliquely downward. There is a predetermined proportion of overlap between multiple photos collected in each circle.

[0012] Furthermore, collect multiple pictures at different heights and different shooting angles of the target position, specifically including:

[0013] Adjust the shooting angle of the gimbal. First, the angle is 0° horizontally, and multiple photos are obtained by shooting a circle horizontally. There is at least 25% of the picture overlap between adjacent photos; keep the position of the aircraft unchanged, tilt the gimbal downward by 30°, and the angle is 30° at this time. Shoot a circle in the same way to obtain multiple materials; keep the position of the aircraft unchanged, then tilt the gimbal downward by 30° again, and the angle is 60° at this time. There is at least 40% of the picture overlap between adjacent photos, and shoot a circle in the same way to obtain multiple materials; finally, take multiple ground photos by vertical downward viewing.

[0014] Furthermore, stitch the multiple collected pictures into a panoramic image through a panoramic image synthesis algorithm, specifically including:

[0015] Detect feature points for each picture and generate feature point descriptions;

[0016] Based on the feature point descriptions of the pictures, perform feature point matching between the pictures;

[0017] Based on the feature point descriptions and matching information of the pictures, calculate the camera parameters;

[0018] Based on the obtained camera parameters, perform projective transformation on the pictures;

[0019] Stitch and reconstruct the pictures after projective transformation into a high-resolution panoramic image.

[0020] Furthermore, detect feature points for each picture and generate feature point descriptions, specifically including:

[0021] For each input picture, use the SIFT algorithm to perform scale-space extreme value detection on the input picture to obtain candidate feature points;

[0022] For the obtained candidate feature points, perform precise positioning by fitting a three-dimensional quadratic function, eliminate points with too low contrast, and use the Hessian matrix to eliminate points with too strong edge effects, and screen out robust feature points;

[0023] For each selected feature point, calculate the gradient magnitude and gradient direction of all pixel points within a circle centered at the feature point with a radius 1.5 times the scale of the Gaussian image where the feature point is located. Then divide the range from 0° to 360° into 36 intervals, and statistically analyze the gradient directions of the pixel points in the form of a histogram according to the corresponding division. The interval with the highest value in the histogram is taken as the main direction of the feature point;

[0024] Finally, construct an image patch around the feature point that is rotationally aligned, i.e., aligned with the main direction of the feature point, and divide the image patch into 4*4 regions. Calculate the gradient histograms in 8 directions: up, down, left, right, upper left, lower left, upper right, and lower right for each region. Finally, generate a 4*4*8 = 128-dimensional vector, which is the descriptor of the feature point.

[0025] Furthermore, based on the description of the feature points of the pictures, perform feature point matching between the pictures, specifically including:

[0026] Based on the description of the feature points of two pictures, use the nearest neighbor matching and RANSAC algorithms to find feature point pairs with geometric consistency through matching and screening:

[0027] Pair up the pictures in the whole group one by one, and a total of n*(n - 1) / 2 pairs of pictures are obtained, where n is the total number of pictures;

[0028] For each pair of pictures, take a feature point in picture A and find two feature points in picture B that are the closest and the second closest to the feature point in A in terms of Euclidean distance. If the ratio of the closest distance to the second closest distance is less than the set threshold, it is considered a valid match and recorded;

[0029] After the matching points are screened, then use the RANSAC algorithm to check the geometric consistency of the matches for further screening: randomly select 4 sample data from the matching dataset and calculate the transformation matrix, then calculate the errors that all feature points appear after using the transformation matrix, and add the feature points with errors less than the set threshold to the inlier set. If the number of elements in the current inlier set is greater than the optimal inlier set, update the optimal inlier set. Repeat the steps of selecting samples, calculating the transformation matrix, calculating errors, and updating the inlier set. Finally, after reaching the preset number of iterations, the algorithm terminates and the screened matching pairs are obtained.

[0030] Furthermore, based on the description of the feature points of the pictures and the matching information, calculate the camera parameters, specifically including:

[0031] Use three tools provided by OpenCV, namely HomographyBasedEstimator, BundleAdjusterRay, and waveCorrect, to estimate the camera parameters of the pictures and perform global optimization:

[0032] First, the feature point description and matching information are used as the input of the HomographyBasedEstimator, from which the algorithm generates a preliminary estimation of the camera translation and rotation matrices between each pair of images;

[0033] Next, the BundleAdjusterRay is used to optimize the estimation results. The algorithm will consider all images and their mutual relationships, continuously iterate and adjust the camera parameters to minimize the reprojection error of all feature points as much as possible, and output a set of optimized camera parameters after the iteration count is exhausted;

[0034] Finally, the waveCorrect is used to perform global smoothing processing on the camera parameters in the horizontal direction to eliminate the drift error.

[0035] Furthermore, based on the obtained camera parameters, a projective transformation is performed on the images, specifically including:

[0036] Use the PyRotationWarper tool provided by OpenCV to process the images:

[0037] First, the surface type and scale factor need to be defined, and the camera intrinsic matrix corresponding to each image is scaled according to the scale factor to adapt to the transformed size;

[0038] Then for each image, the image, the corresponding camera rotation matrix, and the adjusted intrinsic matrix are used as the input of the algorithm. The algorithm will determine the relationship between the image and the surface according to the rotation matrix and the intrinsic matrix, project each pixel in the image onto the surface, and finally remap the pixels on the surface back to the two-dimensional plane to complete the projective transformation.

[0039] Furthermore, the images after projective transformation are stitched to reconstruct a high-resolution panoramic image, specifically including:

[0040] Use the multi-band fusion method to stitch the images. At the same time, in order to reduce the stitching traces and make the obtained result more natural, the ExposureCompensator_GAIN_BLOCKS tool provided by OpenCV is also used to perform exposure compensation on the images:

[0041] For each image, the exposure compensator first divides it into several small regions according to the preset size. Then for each small region, the exposure compensator will calculate the gain value according to the regional brightness characteristics and the brightness difference between the region and other regions. The calculated gain value will be applied to the corresponding region to adjust the regional brightness. Finally, the image with adjusted brightness, together with the corresponding mask and coordinates, is input into the fuser;

[0042] After all the pictures are input, the fuser first repeatedly downsamples and Gaussian blurs each picture to build a Gaussian pyramid, and calculates the Laplacian pyramid from the Gaussian pyramid. Then, image fusion is performed on each layer of the Laplacian pyramid, that is, the Laplacian images at different scales are stitched together with the help of a mask to obtain the Laplacian pyramid of the panoramic image. Starting from the lowest-resolution image at the top of the pyramid, it is upsampled layer by layer and weighted and synthesized with the next layer to finally reconstruct a high-resolution panoramic image.

[0043] According to a second aspect, in an embodiment, an information processing system for remote field survey data is provided. The system includes:

[0044] A data acquisition module for acquiring multiple pictures at different heights and different shooting angles of a target location;

[0045] A panoramic image synthesis module for stitching the acquired multiple pictures into a panoramic image through a panoramic image synthesis algorithm;

[0046] A data management module for uploading the synthesized panoramic image to a pre-constructed field survey data management platform for remote work to implement functions of viewing, managing, and editing the panoramic image, and for associating and linking the panoramic image with the corresponding project electronic document materials.

[0047] According to a third aspect, in an embodiment, an electronic device is provided. The device includes: a processor and a memory;

[0048] The memory is used for storing one or more program instructions;

[0049] The processor is used for running one or more program instructions to execute the steps of an information processing method for remote field survey data as described in any one of the above.

[0050] According to a fourth aspect, in an embodiment, a computer-readable storage medium is provided. A computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of an information processing method for remote field survey data as described in any one of the above are implemented.

[0051] This application provides a method and system for information processing of remote field survey data. The panoramic photos obtained through panoramic photo synthesis technology enrich the form of railway engineering field survey image data, and a data management platform for field survey data based on panoramic pictures is constructed. The integration of field data based on panoramic pictures is realized through the binding document data association function. The present invention creatively applies panoramic photos using panoramic photo synthesis technology to railway engineering field surveys, providing more intuitive and vivid image data for the field site, while solving the problem of scattered field data in railway engineering and improving the efficiency of field data collation and later viewing. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a flowchart of a method for information processing of remote field survey data provided by an embodiment of the present invention;

[0053] Figure 2 It is an interface diagram of panoramic map synthesis software in a method for information processing of remote field survey data provided by an embodiment of the present invention;

[0054] Figure 3 It is an interface diagram of a data management platform for field survey data in a method for information processing of remote field survey data provided by an embodiment of the present invention;

[0055] Figure 4 It is an overall architecture diagram of a system for information processing of remote field survey data provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0056] The present invention will be further described in detail below in conjunction with the accompanying drawings through specific embodiments. Similar elements in different embodiments are labeled with related similar element numbers. In the following embodiments, many details are described to make the present application better understood. However, those skilled in the art can easily recognize that some of the features can be omitted in different situations, or can be replaced by other elements, materials, or methods. In some cases, some operations related to the present application are not shown or described in the specification to avoid the core part of the present application being overwhelmed by excessive description. For those skilled in the art, it is not necessary to describe these related operations in detail, and they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0057] In addition, the features, operations, or characteristics described in the specification can be combined in any appropriate manner to form various embodiments. At the same time, the steps or actions in the method description can also be reordered or adjusted in a manner that is obvious to those skilled in the art. Therefore, the various sequences in the specification and drawings are only for clearly describing a certain embodiment and do not mean a necessary sequence, unless it is stated otherwise that a certain sequence must be followed.

[0058] A method for information processing of remote field survey data provided by the first embodiment of the present invention is mainly divided into two stages: data acquisition and processing, and data management. First, the survey personnel use a drone equipped with a camera to collect photos at three different heights at the same location. At each height, multiple photos are collected by horizontally and obliquely downwardly rotating the camera 360 degrees respectively. Then, the feature points in the images are extracted by the SIFT algorithm from the obtained photos for matching, and multiple two-dimensional photos are stitched into a panoramic image. Finally, the synthesized panoramic image is uploaded to the established field data management platform, where functions such as viewing, managing, and editing of the panoramic image are realized. At the same time, the panoramic image is associated and linked with the corresponding project electronic document materials for unified management. This solution is developed based on the computer web side and can be extended to the mobile Android operating system to achieve the same technical effects. The following will be described in detail in conjunction with Figure 1 for detailed description.

[0059] As Figure 1 shown, in step S100, multiple pictures at different heights and different shooting angles of the target location are collected.

[0060] The above steps specifically include:

[0061] S110, flying the drone equipped with a camera to the designated target location and determining the hovering height according to the height of the on-site building;

[0062] S120, at different hovering heights, adjusting the gimbal angle, and collecting multiple photos by horizontally and obliquely downwardly rotating the camera 360 degrees at different preset angles respectively. There is a predetermined overlap ratio between the multiple photos collected in each circle.

[0063] In this embodiment, first, fly the drone to the designated position, determine the hovering height according to the height of the on-site building, generally hover at 60 - 120 meters, adjust the shooting angle of the pan-tilt head. First, the angle is 0° horizontally, and about 8 - 16 photos are taken horizontally in a circle, with at least 25% of the picture overlap between adjacent photos; keep the position of the aircraft unchanged, tilt the pan-tilt head downward by 30°, at this time the angle is 30°, and about 8 - 16 pieces of footage are taken in a circle in the same way; keep the position of the aircraft unchanged, then tilt the pan-tilt head downward by another 30°, at this time the angle is 60°, with at least 40% of the picture overlap between adjacent photos, and about 8 - 16 pieces of footage are taken in a circle in the same way; finally, take 4 ground photos vertically from above, and rotate each photo by 90°.

[0064] As Figure 1 shown, in step S200, a plurality of collected pictures are stitched into a panoramic image through a panoramic image synthesis algorithm.

[0065] The above steps specifically include:

[0066] S210, detect feature points for each picture and generate feature point descriptions;

[0067] S220, based on the feature point descriptions of the pictures, perform feature point matching between the pictures;

[0068] S230, based on the feature point descriptions of the pictures and the matching information, calculate the camera parameters;

[0069] S240, based on the obtained camera parameters, perform projective transformation on the pictures;

[0070] S250, perform stitching on the pictures after projective transformation to reconstruct a high-resolution panoramic image.

[0071] The specific steps are as follows:

[0072] After the shooting is completed, upload a group of pictures taken at the same position to the self-developed panoramic picture synthesis software (as Figure 2 shown) for synthesis processing. The synthesis is divided into the following five steps: feature point detection, feature point matching, camera parameter calculation, projective transformation, and stitching.

[0073] I) Feature point detection:

[0074] Through the SIFT algorithm, detect feature points for each picture and record the corresponding feature point descriptions.

[0075] For each input image, the SIFT algorithm first performs scale-space extreme value detection, that is, constructs a Difference-of-Gaussian (DoG) pyramid for the image and compares the value of each pixel with its 26 adjacent pixels (8 in the neighborhood, 9 in the upper and lower scale levels). If the pixel value is the largest or the smallest, it is recorded as a candidate feature point. Then, a three-dimensional quadratic function is fitted to accurately locate the obtained candidate feature points, and points with too low contrast are removed. Points with strong edge effects are removed using the Hessian matrix. Finally, robust feature points are selected. For each selected feature point, the algorithm calculates the gradient magnitude and gradient direction of all pixel points within a circle centered at this feature point with a radius of 1.5 times the scale of the Gaussian image where the feature point is located. Then, the range from 0° to 360° is evenly divided into 36 intervals, and the gradient directions of the pixel points are statistically analyzed in the form of a histogram according to this division. The interval with the highest value in the histogram is used as the main direction of this feature point. Finally, the algorithm constructs an image patch around the feature point that is rotationally aligned (aligned with the main direction of the feature point), and divides this image patch into 4*4 regions. Each region calculates the gradient histogram in 8 directions (up, down, left, right, upper left, lower left, upper right, lower right). Finally, a 4*4*8 = 128-dimensional vector is generated, which is the descriptor of this feature point.

[0076] Through the above steps, the program extracts a set of feature points from each image. These feature points have scale invariance, rotation invariance, and a certain degree of illumination change invariance, which can facilitate subsequent feature point matching.

[0077] II) Feature point matching:

[0078] This step uses the nearest neighbor matching and RANSAC algorithms. Based on the feature descriptions of two images, feature point pairs with geometric consistency are found through matching and screening.

[0079] First, pair up all the images in the group. A total of n*(n - 1) / 2 pairs of images can be obtained (n is the total number of images). For each pair of images, take a feature point in image A and find two feature points in image B with the closest Euclidean distance to it. Among these two feature points, if the ratio of the closest distance to the second-closest distance is less than the set threshold, it is considered a valid match and recorded. After the above matching point screening, the RANSAC algorithm is then used to check the geometric consistency of the matches for further screening. Randomly select 4 sample data from the matching dataset and calculate the transformation matrix. Then, calculate the errors that occur for all feature points after using this transformation matrix, and add those with errors less than the set threshold to the inlier set. If the number of elements in the current inlier set is greater than the optimal inlier set, update the optimal inlier set. Repeat the steps of selecting samples, calculating the transformation matrix, calculating errors, and updating the inlier set. Finally, after the iteration count is exhausted, the algorithm terminates and the screened matching pairs are obtained.

[0080] Through the above steps, the program obtains the feature point matching information between each pair of images, and these matching pairs have geometric consistency.

[0081] III) Camera parameter calculation:

[0082] In this step, three tools provided by OpenCV, namely HomographyBasedEstimator, BundleAdjusterRay, and waveCorrect, are used to estimate the camera parameters of the images and perform global optimization.

[0083] First, the feature point descriptions and matching information are used as the input of HomographyBasedEstimator. The algorithm will generate a preliminary estimate of the camera translation and rotation matrices between each pair of images. However, this matrix does not consider global consistency and has errors, and the errors will gradually accumulate as the number of images increases, eventually leading to misalignment or distortion of the stitching result. Therefore, BundleAdjusterRay is used next to optimize the estimation result. The algorithm will consider all images and their mutual relationships, continuously iterate and adjust the camera parameters to minimize the reprojection error of all feature points, and output a set of optimized camera parameters after the iteration count is exhausted. Finally, waveCorrect is used to perform global smoothing processing on the parameters in the horizontal direction to eliminate the drift error.

[0084] Through the above steps, the program obtains a set of accurately calculated camera parameters, which can reduce misalignment, distortion, and blurring problems in stitching and make the stitching result more natural.

[0085] IV) Projection transformation:

[0086] The panoramic view display adopts the method of projecting the images onto a spherical surface and observing from the center of the sphere. To eliminate the perspective difference, it is necessary to perform projection transformation on the images in advance. Here, the PyRotationWarper tool provided by OpenCV is used to process the images.

[0087] First, it is necessary to define the surface type (here it is a spherical surface) and the scale factor, and scale the camera intrinsic matrix corresponding to each image according to the scale factor to adapt to the transformed size. Then, for each image, the image, the corresponding camera rotation matrix, and the adjusted intrinsic matrix are used as the input of the algorithm. The algorithm will determine the relationship between the image and the surface according to the rotation matrix and the intrinsic matrix, project each pixel in the image onto the surface, and finally remap the pixels on the surface back to the two-dimensional plane to complete the projection transformation.

[0088] Through the above steps, the program obtains the images after projection transformation, their corresponding masks, and the coordinates used for alignment in the subsequent stitching process.

[0089] (V) Stitching:

[0090] In this step, the method of multi - band fusion is used to stitch the pictures. At the same time, in order to reduce the stitching marks and make the obtained result more natural, the ExposureCompensator_GAIN_BLOCKS tool provided by OpenCV is also used to perform exposure compensation on the pictures.

[0091] For each picture, the exposure compensator first divides it into several smaller regions according to a preset size. Then, for each region, the exposure compensator calculates the gain value according to the brightness characteristics of this region and the brightness difference between this region and other regions. The calculated gain value will be applied to the corresponding region to adjust the brightness of this region. Finally, the picture with adjusted brightness, together with its corresponding mask and coordinates, will be input into the fuser. After all pictures are input, the fuser will first perform downsampling and Gaussian blur on each picture repeatedly to build its corresponding Gaussian pyramid, and calculate the Laplacian pyramid from the Gaussian pyramid. Then, image fusion is performed on each layer of the Laplacian pyramid, that is, the Laplacian images at different scales are stitched together with the help of the mask to obtain the Laplacian pyramid of the panoramic image. Starting from the lowest - resolution image (i.e., the top of the pyramid), it is upsampled layer by layer and weighted - synthesized with the next layer, and finally a high - resolution panoramic image is reconstructed.

[0092] At this point, the program has completed the synthesis of the captured pictures and obtained the required panoramic image.

[0093] As Figure 1 shown, in step S300, the synthesized panoramic image is uploaded to the pre - constructed field survey data management platform to implement the functions of viewing, managing, and editing the panoramic image, as well as associating and linking the panoramic image with the corresponding project electronic document materials.

[0094] Specifically, finally, the panoramic image is uploaded to the data management platform (as Figure 3 shown) for functions such as classified viewing, annotation editing, and document association of the panoramic image.

[0095] Corresponding to the above - disclosed method for information - based processing of remote field survey data, an embodiment of the present invention also discloses a system for information - based processing of remote field survey data, as Figure 4 shown, which specifically includes:

[0096] A data acquisition module, which is used to acquire multiple pictures at different heights and different shooting angles of the target location;

[0097] A panoramic image synthesis module, which is used to stitch the multiple acquired pictures into a panoramic image through a panoramic image synthesis algorithm;

[0098] A data management module, which is used to upload the synthesized panoramic image to a pre-constructed field survey data management platform, implement functions such as viewing, managing, and editing the panoramic image, and link the panoramic image with the corresponding project electronic document materials.

[0099] It should be noted that the detailed description of a remote field survey data informatization processing system provided by an embodiment of the present invention can refer to the relevant description of a remote field survey data informatization processing method provided by an embodiment of the present application, and will not be elaborated here.

[0100] In addition, an embodiment of the present invention further provides an electronic device, which includes: a processor and a memory; the memory is used to store one or more program instructions; the processor is used to run one or more program instructions to execute the steps of a remote field survey data informatization processing method as described in any one of the above.

[0101] It should be noted that the detailed description of an electronic device provided by an embodiment of the present invention can refer to the relevant description of a remote field survey data informatization processing method provided by an embodiment of the present application, and will not be elaborated here.

[0102] In addition, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of a remote field survey data informatization processing method as described in any one of the above.

[0103] It should be noted that the detailed description of a computer-readable storage medium provided by an embodiment of the present invention can refer to the relevant description of a remote field survey data informatization processing method provided by an embodiment of the present application, and will not be elaborated here.

[0104] Those skilled in the art can understand that all or part of the functions of the various methods in the above embodiments can be implemented in a hardware manner or in a computer program manner. When all or part of the functions in the above embodiments are implemented in a computer program manner, the program can be stored in a computer-readable storage medium, and the storage medium can include: read-only memory, random access memory, magnetic disk, optical disk, hard disk, etc. The above functions can be realized by a computer executing the program. For example, the program is stored in the memory of the device, and when the processor executes the program in the memory, the above all or part of the functions can be realized. In addition, when all or part of the functions in the above embodiments are implemented in a computer program manner, the program can also be stored in a storage medium such as a server, another computer, magnetic disk, optical disk, flash drive or mobile hard disk, and saved to the memory of the local device by downloading or copying, or the system of the local device is updated. When the processor executes the program in the memory, all or part of the functions in the above embodiments can be realized.

[0105] The above uses specific examples to elaborate on the present invention, which is only used to help understand the present invention and is not intended to limit the present invention. For those skilled in the art of the present invention, according to the idea of the present invention, several simple deductions, deformations or substitutions can also be made.

Claims

1. A method for informatization processing of remote field survey data, characterized in that: The method comprises: Collect multiple pictures of the target at different heights and shooting angles; The collected multiple pictures are stitched into a panoramic image through a panoramic image synthesis algorithm; Upload the synthesized panoramic image to the pre-built field survey data management platform to realize the viewing, management and editing functions of the panoramic image, and link the panoramic image with the corresponding project electronic document information.

2. A remote field survey data information processing method as claimed in claim 1, characterized in that: Collect multiple pictures of the target location at different heights and shooting angles, including: Fly the drone carrying the camera to the designated target location and determine the hovering height based on the height of the buildings on site; At different hovering heights, adjust the gimbal angle, and collect multiple photos in a 360-degree circle at different preset camera angles, horizontally and diagonally downward. There is a predetermined ratio of overlap between the multiple photos collected in each circle.

3. A remote field survey data information processing method as claimed in claim 2, characterized in that: Collect multiple pictures of the target location at different heights and shooting angles, including: Adjust the gimbal shooting angle. First, the angle is 0° horizontally. Shoot in a circle horizontally to get multiple photos. At least 25% of the pictures between adjacent photos overlap. Keep the aircraft position unchanged, move the gimbal downward 30°, now the angle is 30°, and shoot in a circle using the same method to get multiple materials. Keep the aircraft position unchanged, move the gimbal downward 30°, now the angle is 60°, and at least 40% of the pictures between adjacent photos overlap. Shoot in a circle using the same method to get multiple materials. Finally, look down vertically to take multiple photos of the ground.

4. The method for informatization processing of remote field survey data according to claim 1, characterized in that: The collected multiple pictures are stitched into a panoramic image through a panoramic image synthesis algorithm, which includes: Detect feature points for each image and generate feature point descriptions; Based on the feature point description of the picture, the feature points between pictures are matched; Calculate camera parameters based on the feature point description and matching information of the image; Based on the obtained camera parameters, the image is projected and transformed; The projected images are stitched together to reconstruct a high-resolution panorama.

5. The method for informatization processing of remote field survey data according to claim 4, characterized in that: Perform feature point detection on each image and generate feature point descriptions, including: For each input image, the SIFT algorithm is used to detect the extreme values ​​in the scale space of the input image to obtain candidate feature points; For the candidate feature points obtained, the three-dimensional quadratic function is fitted to accurately locate and eliminate points with too low contrast, and the Hessian matrix is ​​used to eliminate points with too strong edge effects to screen out robust feature points; For each feature point selected, calculate the gradient amplitude and gradient direction of all pixels in a circle with the feature point as the center and 1.5 times the scale of the Gaussian image where the feature point is located as the radius, then divide 0° to 360° into 36 intervals, and count the gradient directions of the pixels in the form of a histogram according to the corresponding divisions. The interval with the highest value in the histogram is taken as the main direction of the feature point; Finally, a rotationally aligned image block is constructed around the feature point, i.e., the image block is aligned with the main direction of the feature point, and the image block is divided into 4*4 regions. The gradient histograms in 8 directions of up, down, left, right, upper left, lower left, upper right, and lower right are calculated for each region, and finally a 4*4*8=128-dimensional vector is generated, which is the descriptor of the feature point.

6. A remote field survey data information processing method as claimed in claim 4, characterized in that: Based on the feature point description of the image, feature point matching is performed between images, including: Based on the feature point description of the two images, the nearest neighbor matching and RANSAC algorithms are used to find feature point pairs with geometric consistency through matching and screening: Pair the entire set of pictures in pairs, and get n*(n-1) / 2 pairs of pictures, where n is the total number of pictures; For each pair of images, take a feature point in image A and find two feature points in image B with the closest and second closest Euclidean distance to the feature point in A. If the closest distance divided by the second closest distance is less than the set threshold, it is considered a valid match and recorded; After the matching points are screened, the RANSAC algorithm is used to check the geometric consistency of the matching for further screening: 4 sample data are randomly selected from the matching data set and the transformation matrix is ​​calculated. Then, the errors of all feature points after using the transformation matrix are calculated, and those with errors less than the set threshold are added to the inliers set. If the number of elements in the current inliers set is greater than the optimal inliers set, the optimal inliers set is updated, and the steps of selecting samples, calculating the transformation matrix, calculating the error and updating the inliers set are repeated. Finally, after reaching the preset number of iterations, the algorithm terminates and the screened matching pairs are obtained.

7. The method for informatization processing of remote field survey data according to claim 4, characterized in that: Based on the feature point description and matching information of the image, the camera parameters are calculated, including: Use the three tools provided by OpenCV, HomographyBasedEstimator, BundleAdjusterRay and waveCorrect, to estimate the camera parameters and globally optimize the image: First, the feature point description and matching information are used as input to the HomographyBasedEstimator, and the algorithm generates a preliminary estimate of the camera translation and rotation matrix between each pair of images; Next, use BundleAdjusterRay to optimize the estimation results. The algorithm will consider all images and their relationships, iteratively adjust the camera parameters, minimize the reprojection error of all feature points, and output a set of optimized camera parameters after the number of iterations is exhausted; Finally, waveCorrect is used to perform global smoothing of the camera parameters in the horizontal direction to eliminate drift errors.

8. The method for informatization processing of remote field survey data according to claim 4, characterized in that: Based on the obtained camera parameters, the image is projected and transformed, including: Use the PyRotationWarper tool provided by OpenCV to process the image: First, you need to define the surface type and scale factor, and scale the camera intrinsic parameter matrix corresponding to each image according to the scale factor to adapt to the transformed size; Then, for each picture, the picture and the corresponding camera rotation matrix and the adjusted intrinsic parameter matrix are used as inputs to the algorithm. The algorithm determines the relationship between the picture and the surface based on the rotation matrix and the intrinsic parameter matrix, projects each pixel in the picture onto the surface, and finally remaps the pixels on the surface back to the two-dimensional plane to complete the projection transformation.

9. The method for informatization processing of remote field survey data according to claim 4, characterized in that: The projected images are stitched together to reconstruct a high-resolution panoramic image, including: The multi-band fusion method is used to stitch the pictures. At the same time, in order to reduce the stitching traces and make the result more natural, the ExposureCompensator_GAIN_BLOCKS tool provided by OpenCV is also used to compensate the exposure of the pictures: For each image, the exposure compensator first divides it into several small areas according to the preset size. Then, for each small area, the exposure compensator calculates the gain value according to the brightness characteristics of the area and the brightness difference between the area and other areas. The calculated gain value is applied to the corresponding area to adjust the brightness of the area. Finally, the image with adjusted brightness is input into the fusion device together with the corresponding mask and coordinates. After all images are input, the fuser will first repeatedly downsample and Gaussian blur each image to build a Gaussian pyramid, and then calculate the Laplacian pyramid from the Gaussian pyramid. Then, image fusion is performed on each layer of the Laplacian pyramid, that is, the Laplacian images of different scales are stitched together with the help of masks to obtain the Laplacian pyramid of the panoramic image. Starting with the lowest resolution image at the top of the pyramid, it is upsampled layer by layer and weightedly synthesized with the next layer to finally reconstruct a high-resolution panorama.

10. A remote field survey data information processing system, characterized in that: The system comprises: The data acquisition module is used to collect multiple pictures of the target position at different heights and shooting angles; A panoramic image synthesis module is used to stitch multiple collected pictures into a panoramic image through a panoramic image synthesis algorithm; The data management module is used to upload the synthesized panoramic images to the pre-built field survey data management platform, realize the viewing, management and editing functions of the panoramic images, and associate and link the panoramic images with the corresponding project electronic document materials.

Images