A method, system, terminal, and storage medium for shallow water depth estimation based on asynchronous observation and scan angle correction.

Abstract

This invention relates to the field of laser signal processing technology, and discloses a method, system, terminal, and storage medium for shallow water depth estimation based on asynchronous observation and scanning angle correction. The method includes: performing multiple asynchronous observations of a land target area for error optimization processing, then performing asynchronous observations of an underwater target area and extracting multiple key point pairs to construct a sparse correspondence; using the sparse correspondence to construct multiple distance values, and further selecting multiple initial key point pairs with consistent distances; performing structural consistency screening on the initial key point pairs and constructing difference values ​​to optimize the estimated water depth value. This invention is based on the spatial detection principle and point cloud solution method of an airborne single-band depth sounding system, avoiding the error influence between multi-source heterogeneous data detection results from multi-system integration, and effectively improving the ground detection accuracy and integrated land-water detection capability of the airborne blue-green lidar depth sounding system.

Description

Technical Field

[0001] This invention relates to the field of laser signal processing technology, and in particular to a method, system, terminal, and computer-readable storage medium for estimating shallow water depth based on asynchronous observation and scanning angle correction. Background Technology

[0002] Current nearshore shallow water topographic detection technology is showing a trend of multimodal integration, mainly relying on airborne lidar water depth detection technology, which in turn depends on the processing and analysis of blue-green laser echo signals.

[0003] However, in practical engineering applications, airborne lidar water depth detection technology, which relies on in-situ water level observation or multi-sensor integrated water surface detection, is often limited by factors such as the difficulty of deploying in-situ observation equipment, system integration conditions, and load constraints, resulting in significant errors in water depth estimation.

[0004] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0005] The main objective of this invention is to provide a shallow water depth estimation method, system, terminal, and computer-readable storage medium based on asynchronous observation and scanning angle correction. This invention aims to solve the problem that existing water depth estimation techniques based on single-system scanning detection suffer from errors between multi-source heterogeneous data detection results when integrating multiple systems, resulting in large errors in the water and land detection results.

[0006] To achieve the above objectives, this invention provides a shallow water depth estimation method based on asynchronous observation and scan angle correction. The shallow water depth estimation method based on asynchronous observation and scan angle correction includes the following steps:

[0007] Multiple asynchronous observations are performed on the land target area to obtain multiple pairs of corresponding points, and an optimization objective function is constructed based on the pairs of corresponding points;

[0008] The results of all asynchronous observations are iteratively optimized according to the aforementioned objective function to obtain the corrected scan angle error.

[0009] Asynchronous observation of the underwater target area is performed to obtain the first underwater point cloud and the second underwater point cloud. The first underwater point cloud and the second underwater point cloud are divided into multiple super voxel regions, and multiple key point pairs are extracted to construct a sparse correspondence relationship.

[0010] Based on the sparse correspondence, a distance value is constructed for each key point pair. All key point pairs are filtered based on all the distance values ​​to obtain multiple initial key point pairs with consistent distances.

[0011] Construct neighboring point sets for the first and second key points in each initial key point pair, and construct difference values ​​based on the two neighboring point sets.

[0012] An optimization function is constructed using all the aforementioned differences, and the water depth of the underwater target area is determined based on the optimization function.

[0013] In this invention, multiple asynchronous observations are performed on a land target area to obtain multiple pairs of corresponding points, and an optimization objective function is constructed based on these pairs. The results of all asynchronous observations are iteratively optimized using the optimization objective function to obtain a corrected scanning angle error. Asynchronous observations are also performed on an underwater target area to obtain a first underwater point cloud and a second underwater point cloud. These two underwater point clouds are divided into multiple supervoxel regions, and multiple keypoint pairs are extracted to construct a sparse correspondence. A distance value is constructed for each keypoint pair based on the sparse correspondence. All keypoint pairs are filtered based on all distance values ​​to obtain multiple initial keypoint pairs with consistent distances. Neighborhood sets of the first and second keypoints in each initial keypoint pair are constructed, and difference values ​​are constructed based on the two neighborhood sets. An optimization function is constructed using all the difference values, and the water depth of the underwater target area is determined based on the optimization function. This invention is based on the space detection principle and point cloud calculation method of airborne single-band depth sounding system, avoiding the error influence between multi-source heterogeneous data detection results such as multi-system integration, and effectively improving the ground detection accuracy and water-land integrated detection capability of airborne blue-green lidar depth sounding system. Attached Figure Description

[0014] Figure 1 This is a flowchart of a preferred embodiment of the shallow water depth estimation method based on asynchronous observation and scanning angle correction of the present invention;

[0015] Figure 2 This is a data processing flowchart of a preferred embodiment of the shallow water depth estimation method based on asynchronous observation and scanning angle correction of the present invention;

[0016] Figure 3 This is a trajectory diagram of an airborne blue-green laser heading scan point cloud, representing a preferred embodiment of the shallow water depth estimation method based on asynchronous observation and scanning angle correction of the present invention.

[0017] Figure 4 This is a trajectory diagram of a circular scan of a preferred embodiment of the shallow water depth estimation method based on asynchronous observation and scan angle correction of the present invention;

[0018] Figure 5 This is a trajectory diagram of an oval scan, representing a preferred embodiment of the shallow water depth estimation method based on asynchronous observation and scan angle correction of the present invention.

[0019] Figure 6 This is a diagram illustrating the scanning angle error correction process of a preferred embodiment of the shallow water depth estimation method based on asynchronous observation and scanning angle correction of the present invention.

[0020] Figure 7 This is a point cloud processing effect diagram after error correction, representing a preferred embodiment of the shallow water depth estimation method based on asynchronous observation and scanning angle correction of the present invention.

[0021] Figure 8 This is a schematic diagram of the geometric relationship of land and water exploration in a preferred embodiment of the shallow water depth estimation method based on asynchronous observation and scanning angle correction of the present invention.

[0022] Figure 9 This is a flowchart of point-to-point matching of underwater point clouds, representing a preferred embodiment of the shallow water depth estimation method based on asynchronous observation and scanning angle correction of the present invention.

[0023] Figure 10 This is a schematic diagram of an integrated land and water point cloud representing a preferred embodiment of the shallow water depth estimation method based on asynchronous observation and scanning angle correction of the present invention.

[0024] Figure 11 This is a schematic diagram of the forward scan point cloud during a rotating scan within the same flight strip, representing a preferred embodiment of the shallow water depth estimation method based on asynchronous observation and scan angle correction of the present invention.

[0025] Figure 12 This is a schematic diagram of the semi-circular point cloud after conical scanning of a preferred embodiment of the shallow water depth estimation method based on asynchronous observation and scanning angle correction of the present invention;

[0026] Figure 13 This is a comparison chart of observational biases of a land-water system at the same location, based on a preferred embodiment of the shallow water depth estimation method of the present invention, which is based on asynchronous observation and scanning angle correction.

[0027] Figure 14 This is a structural diagram of a preferred embodiment of the shallow water depth estimation system based on asynchronous observation and scanning angle correction of the present invention;

[0028] Figure 15 This is a structural diagram of a preferred embodiment of the terminal of the present invention. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0030] The preferred embodiment of the present invention describes a shallow water depth estimation method based on asynchronous observation and scan angle correction, such as... Figure 1 As shown, the shallow water depth estimation method based on asynchronous observation and scan angle correction includes the following steps:

[0031] Step S10: Perform multiple asynchronous observations on the land target area to obtain multiple pairs of corresponding points, and construct an optimized objective function based on the pairs of corresponding points.

[0032] In asynchronous observations, the repeated observation results of the airborne blue-green lidar depth sounding system for the same area often differ. Considering the high speed of the airborne platform and the small spatial range of the local overlapping area, the difference in observation results caused by instantaneous environmental changes can be ignored. Therefore, the main reason for the deviation in the asynchronous observation results of underwater three-dimensional terrain comes from the error of the incident direction and the influence of the propagation direction of the laser system during cross-medium propagation.

[0033] Therefore, in embodiments of the present invention, such as Figure 2 As shown, a regional water depth estimation method for asynchronous observation bias in a single-band depth sounding lidar system is presented. This method addresses situations where the water depth is shallow or the surface laser echo signal intensity is low, and the surface and bottom are superimposed and difficult to extract. It constructs an underwater local point cloud position deviation estimation model, using water depth as a parameter, and employs a nonlinear optimization method for numerical solution, thereby achieving local water depth estimation. This provides an effective technical solution for airborne blue-green lidar depth sounding systems in near-shore extremely shallow water areas.

[0034] Specifically, the land target area is scanned asynchronously multiple times to obtain multiple first land observation point clouds and multiple second land observation point clouds; based on each first land observation point cloud and each second land observation point cloud, multiple pairs of points with the same name are constructed; based on all the pairs of points with the same name, the spatial positional deviation of the asynchronous observation is constructed:

[0035] ;

[0036] in, This indicates the first scan point of asynchronous land observation. This indicates the second scan point of asynchronous land observation. Indicates the first laser emission point for asynchronous observation on land. This indicates the second laser emission point for asynchronous observation on land. Representing vector symbols, express arrive The vector, express arrive The vector, express arrive The vector, , and They represent x-coordinate, y-coordinate, and y-coordinate , and They represent x-coordinate, y-coordinate, and y-coordinate express arrive distance, express arrive distance, and These represent the first spatial incident angle and the second spatial incident angle of the asynchronous land observation, respectively. and These represent the first spatial azimuth and the second spatial azimuth of the asynchronous land observation, respectively. Indicates the scanning angle error;

[0037] Using the aforementioned spatial position deviation, an optimization objective function is constructed:

[0038] ;

[0039] in, This indicates the number of pairs of nodes with the same name. Indicates to index, This represents the objective function to be optimized. This indicates taking the minimum value.

[0040] In embodiments of the present invention, asynchronous observation is performed using an asynchronous scanning observation model based on a system rotational scanning structure, such as... Figure 3 As shown, a circular or near-circular optomechanical scanning strategy is employed to perform two asynchronous scans of the same location along the flight direction, forming an asynchronous observation point cloud of the same scanned area within the flight path. This point cloud consists of a forward scan point cloud and a backward scan point cloud. The specific scanning trajectory is as follows: Figure 4 and Figure 5 As shown, the horizontal axis X and the vertical axis Y represent the size of the point cloud scan on the horizontal and vertical axes, respectively. Although the overlapping area of ​​the asynchronously scanned point cloud is scanned twice, the spatial position of the detected target should be consistent in the scanning arc segments in the forward and backward directions. However, due to the influence of system scanning direction error and laser refraction under different water depth conditions in cross-medium observation, the three-dimensional scanning result of the asynchronously scanned area usually shows a spatial position deviation of the corresponding points obtained from the two scans.

[0041] Considering the asynchronous scanning observation point cloud in the land area, there is no refraction change in cross-medium detection. Therefore, the scanning angle error of the system is estimated by the spatial position deviation of the asynchronous observation point cloud on land, and the scanning angle error is corrected for the underwater detection results.

[0042] Specifically, as mentioned above, the formula for constructing the spatial position deviation of asynchronous observations is the scanning angle error correction model of the airborne blue-green lidar land target spatial position system. Since the spatial positions of the two asynchronous observations are relatively close, their theoretical values ​​should be at the same point, i.e. Approaching 0, therefore, the pair can be... The estimation is transformed into the spatial position deviation of the asynchronous detection results of the target point. The optimization problem is minimized, thus constructing an optimization objective function; when the asynchronous observation deviation value tends to be minimized, the corresponding scanning angle error is the optimal value.

[0043] Step S20: Iteratively optimize the results of all asynchronous observations according to the optimization objective function to obtain the corrected scan angle error.

[0044] Specifically, the scanning angle error of each pair of corresponding points is defined as an individual, and multiple individuals are randomly selected as an initial population; the initial population is screened using the optimization objective function to obtain multiple excellent individuals whose scanning angle errors are less than a preset scanning angle error; a crossover operation is performed on all the excellent individuals to generate multiple offspring individuals, and a mutation operation is performed on all the offspring individuals to generate an offspring population; the offspring population is iterated until a target offspring population is obtained, and the corrected scanning angle error is screened from the target offspring population.

[0045] Among them, such as Figure 6 As shown, Matlab (Matrix Laboratory) is used to perform scan angle error correction based on a genetic algorithm, employing a real-number encoding method. Let each individual in the genetic algorithm represent a specific incident angle value. Then, a set of incident angle values ​​is randomly generated as the initial population (the size of the initial population can be determined according to the user's needs, the complexity of the scanning problem, or computing resources to improve the applicability of the method described in this invention). A crossover operation is performed on the selected parent individuals to generate new offspring individuals (in this embodiment of the invention, the crossover probability can be set to 0.7, but not limited to this). Then, a mutation operation is performed on the offspring individuals (in this embodiment of the invention, the mutation probability can be set to 0.5, but not limited to this) to increase the diversity of the population (wherein, the mutation operation can be adding a small random number to the incident angle value or making random adjustments within a certain range). The above selection, crossover, and mutation steps are repeated until a predetermined number of iterations is reached or the fitness function value converges to below a certain threshold. Finally, the individual with the highest fitness is selected from the final population as the optimal solution, i.e., the corrected scanning angle error.

[0046] In this embodiment of the invention, the maximum number of iterations is 110, and the final scan angle error is 1.35°. This is then used in the original data processing correction parameters to regenerate the point cloud data. Figure 7 As shown, by visualizing the regenerated land laser point cloud after correction, it can be seen that the overlap of the front and rear semi-circular point clouds in the national highway point cloud section is good. Distance analysis of key point pairs in the point cloud after considering the correction error processing is performed and compared with the uncorrected point pairs. The results show that the method of the present invention uses the matching deviation of the land laser point cloud to estimate the scanning angle error of the system, thereby eliminating the laser propagation direction deviation caused by system placement deviation or structural system error in lidar water depth detection, thus effectively improving the accuracy of the system scanning detection results and laying the foundation for system water depth estimation.

[0047] Step S30: Asynchronously observe the underwater target area to obtain the first underwater point cloud and the second underwater point cloud. Divide the first underwater point cloud and the second underwater point cloud into multiple supervoxel regions and extract multiple key point pairs to construct a sparse correspondence relationship.

[0048] In another embodiment of the present invention, the supervoxel region includes a first supervoxel region and a second supervoxel region. Because the surface echo signal intensity is relatively weak in shallow water areas and merges with the bottom echo signal, it is difficult to obtain the laser's arrival time at the water surface. Therefore, determining the propagation time of the blue-green laser in the water after asynchronous laser observation always occurs from different directions at the water surface becomes a key issue in this embodiment of the present invention. Figure 8 As shown, the increment from the detector's spatial position to its underwater detection position can be expressed as:

[0049] ;

[0050] ;

[0051] in, and These represent the first underwater detection position and the second underwater detection position, respectively. and Let represent the vector from the first laser emission point to the first underwater detection position and the vector from the second laser emission point to the second underwater detection position, respectively. and Let represent the vector from the first laser emission point to the incident position on the first water surface and the vector from the second laser emission point to the incident position on the second water surface, respectively. and Let represent the vector from the first water surface incident position to the first underwater detection position and the vector from the second water surface incident position to the second underwater detection position, respectively. This indicates the speed at which laser light travels through the air. and These represent the first and second propagation times of the underwater echo signal obtained from asynchronous laser observation, respectively. and They represent, The refractive index represents the refractive index of a laser beam as it propagates across a medium, such as air or water. and They represent and The corresponding angle of refraction after incident on the water surface.

[0052] Furthermore, the asynchronous observation model for underwater target points can be expressed as:

[0053] ;

[0054] ;

[0055] Among these, the water depth obtained from asynchronous observations should have a high degree of consistency. When the errors in the various parameters during the observations are small, the same target location... Asynchronous observation bias The influence should be relatively small; as can be seen from the observation model, the main factors affecting the spatial position of underwater targets in depth-sounding lidar include the time position of the signal echo from the water surface and the bottom, the system scanning angle and scanning azimuth, and the water refractive index. Among these, the echo signal time mainly depends on the accuracy of the echo signal detection and analysis, the system scanning angle is mainly related to the installation accuracy of the scanning mechanical structure, and the system scanning azimuth is mainly related to the synchronization of the scanning mechanical structure and the signal transmission and reception time, and has a high accuracy.

[0056] Furthermore, due to the relatively close incident spatial position and high flight speed, the deviations in the dynamic water surface and its optical properties (such as refractive index) can be considered negligible. Therefore, the system scanning angle and the time position obtained from the analysis of the water surface and bottom echo signals are the main factors affecting the underwater target detection accuracy of the airborne blue-green lidar depth sounding system. In order to achieve water depth estimation in extremely shallow water areas, based on the scanning angle deviation of the land asynchronous observation point cloud estimation system, a step is implemented to estimate the shallow water depth based on the underwater laser point cloud.

[0057] Specifically, asynchronous observation of the underwater target area yields a first underwater point cloud and a second underwater point cloud. These are then uniformly segmented according to minimizing the feature distance between super-voxels, resulting in multiple first and second super-voxel regions. Based on the feature distance, corresponding feature labels are added to all first and second super-voxel regions. The centroids of all first and second super-voxel regions are determined, and multiple first keypoints corresponding to the first underwater point cloud and multiple second keypoints corresponding to the second underwater point cloud are selected based on these centroids.

[0058] ;

[0059] in, Represents the feature distance. Indicates the echo signal strength characteristic label, express The weight, Labels representing the difference in point cloud distances. express The weight, Indicates the feature label of the normal direction of the point cloud. express The weight, The distance between the centroid and the first or second key point is represented; multiple second key points similar to the feature labels of the first key points in the multiple first supervoxel regions are identified in the second underwater point cloud, and key point pairs are constructed based on the first and second key points; a sparse correspondence between the first and second key points in each key point pair is constructed:

[0060] ;

[0061] in, Symbols representing similarity calculations. Feature description representing the first key point, The feature description representing the second key point, This represents the number of feature dimensions in a keypoint pair. The first key point Dimensional features, The second key point Dimensional features, Indicates to The index.

[0062] Among these factors, due to the non-uniform optical properties of water bodies, rapid changes in the system's scanning incident direction, and dynamic water surface incident deviations, the spatial structure and positional deviations between the asynchronous observation point clouds generated through calculation are difficult to describe using simple rigid transformations; therefore, as... Figure 9 As shown, the bending directions of the strip-shaped targets are inconsistent, so a non-rigid point cloud matching method should be studied for this part.

[0063] In this process, the first and second underwater point clouds are jointly segmented using a method of uniformly selecting seed points for growth. The super-voxel features can be selected from signal intensity, terrain features, and spatial distance, etc. The search is performed in a certain order within the seed neighborhood, and the segmentation develops along the voxel with the smallest feature distance from the super-voxel, with all seed points growing simultaneously. The asynchronous point cloud is divided into multiple super-voxel regions, and corresponding feature labels are added to regions with similarity in the segmentation results of different point clouds.

[0064] Furthermore, by combining information such as the spatial distribution of laser points and echo intensity, feature descriptions of key points in the first and second underwater point clouds are calculated (i.e., and Feature descriptors that effectively reflect the spatial structure and neighborhood relationships of point clouds can be selected, such as Fast Point Feature Histograms (FPFH), Heat Kernel Signature (HKS), and 3D Shape Context (3DSC) descriptors. The top few points with the highest similarity to the keypoint feature descriptions in the first underwater point cloud are identified from all points in the second underwater point cloud, thus establishing the initial sparse correspondence between keypoints.

[0065] Step S40: Construct the distance value of each key point pair according to the sparse correspondence, and filter all key point pairs according to all the distance values ​​to obtain multiple initial key point pairs with consistent distances.

[0066] Specifically, based on the sparse correspondence, a first correspondence and a second correspondence are determined in each keypoint pair:

[0067] ;

[0068] ;

[0069] in, Indicates the first correspondence. This indicates the second correspondence. Indicates feature dimension as The first key point, Indicates feature dimension as The second key point, Indicates feature dimension as The first key point, Indicates feature dimension as The second key point, This represents the index indicating the number of feature dimensions for the first keypoint. The index represents the number of feature dimensions for the second keypoint; the distance value corresponding to the keypoint pair is calculated based on each of the first and second correspondences:

[0070] ;

[0071] in, This represents the distance value between keypoint pairs. Distance calculation symbol, This indicates taking the minimum value; the distance value of each keypoint pair is compared with the distance threshold. If the distance value exceeds the distance threshold, the keypoint pair is deleted, resulting in multiple initial keypoint pairs with consistent distances.

[0072] Specifically, by performing supervoxel segmentation and labeling on two point clouds, it is determined whether the local feature labels of the two points associated with each point pair in the initial sparse correspondence are consistent after supervoxel segmentation. Point pairs with consistent regional features are retained, narrowing the matching range of key points to within the supervoxel space and further reducing erroneous matching relationships. Distance consistency and structural consistency assessments are performed on the points associated with each matching pair to eliminate a large number of erroneous point pairs remaining in the initial sparse correspondence.

[0073] Step S50: Construct a set of neighboring points for the first and second key points in each initial key point pair, and construct a difference value based on the two sets of neighboring points.

[0074] Specifically, for each pair of initial keypoints, a corresponding set of neighboring points is constructed for the first keypoint and the second keypoint:

[0075] ;

[0076] in, The index representing the number of initial keypoint pairs. This indicates the number of initial keypoint pairs. Indicates the first The first or second key point in each initial keypoint pair; constructing a linear representation of all elements in each nearest neighbor set:

[0077] ;

[0078] ;

[0079] in, Indicates the first The weight coefficient of the first key point in an initial key point pair Indicates the first The first key point in an initial key point pair Indicates the first The weight coefficient of the second keypoint in the initial keypoint pair Indicates the first The second keypoint in each initial keypoint pair; construct the difference value for the corresponding initial keypoint pair based on all the linear representations:

[0080] ;

[0081] in, Indicates the difference value.

[0082] The structural consistency assessment of key points is based on the local structural information of the two key points associated with the correspondence, which can obtain relatively uniform matching pairs across the entire point cloud. Assuming that the data is linear within a small local range, the two key points associated with each matching pair in the correspondence are represented by a linear combination of their adjacent points within their local region.

[0083] Step S60: Construct an optimization function using all the said difference values, and determine the water depth value of the underwater target area based on the optimization function.

[0084] Specifically, the minimum difference value is determined among all the difference values, and the initial keypoint pair corresponding to the minimum difference value is defined as the target keypoint pair. The remaining initial keypoint pairs are then constructed into keypoint matching pairs.

[0085] ;

[0086] in, Indicates keypoint matching pairs, , and These represent the first keypoints of the first, second, and third initial keypoint pairs in the keypoint matching pair, respectively. , and These represent the second keypoints of the first, second, and third initial keypoint pairs in the keypoint matching pair, respectively. and Let represent the first underwater point cloud and the second underwater point cloud, respectively; based on the key point matching pairs, construct an optimization function for the underwater asynchronous observation bias of the target key point pairs:

[0087] ;

[0088] in, Represents the optimization function. express arrive The vector, This indicates the number of initial keypoint pairs in the keypoint matching pair. express index, This indicates the first key point of the target key point pair. The second key point represents the target key point pair; the optimization function is minimized to estimate the water depth of the underwater target area:

[0089] ;

[0090] in, Indicates the water depth value. Indicates the distance threshold. This indicates the laser incident direction for underwater detection. This represents a constant, with a value of 1 or 2. This indicates the direction of the first laser incident point for underwater detection. This indicates the incident direction of the second laser for underwater detection. Indicates transpose. , and They represent x-coordinate, y-coordinate, and y-coordinate , and They represent x-coordinate, y-coordinate, and y-coordinate This represents the minimum propagation time of a laser signal in water. and These respectively represent laser signals in terms of and The propagation time in water at the time of incident.

[0091] Among them, the point pair with the smallest difference value is selected as the corresponding key point pair in the two point clouds based on the above difference values. The above point pair filtering results are merged, and duplicate matches are removed. The remaining key point pairs are used as key point matching pairs in the final observation error correction model. The pair with the highest feature similarity is selected. Construct an asynchronous observation bias optimization objective function using points of the same name.

[0092] In this embodiment of the invention, the underwater laser propagation time can be estimated using a linear or nonlinear parameter estimation method with the above formula as the objective function. Based on the spatial detection principle and point cloud calculation method of the airborne single-band depth sounding system, it breaks through the water depth estimation technology based on single-system scanning detection, avoids the error influence between multi-source heterogeneous data detection results such as multi-system integration, and can effectively improve the ground detection accuracy and integrated land and water detection capability of the airborne blue-green band airborne blue-green lidar depth sounding system. It has the advantages of simple system, easy operation, and high accuracy.

[0093] Furthermore, in another embodiment of the present invention, the processing effect of the self-developed airborne blue-green lidar depth sounding system in a mixed water-land area is illustrated: the detected lake bottom sediment is mainly silt deposits, which have low reflectivity to blue-green lasers, thus affecting the intensity of laser echoes. The water is also relatively turbid, with an average depth of approximately 0.7m to 1m. Based on a comprehensive analysis of the bottom sediment, water quality, and water depth, the airborne blue-green lidar depth sounding system can meet the requirements for land area measurement work. In this embodiment of the present invention, multiple calibration points are obtained using unmanned surface vessel single-beam measurement, and the system's depth detection accuracy is improved by calibrating the regional water depth measurement results.

[0094] Among them, such as Figure 10 The diagram illustrates the integrated land and water point cloud situation in this embodiment. After initial processing of the acquired data, the system obtains raw point cloud data. Following preprocessing and point cloud denoising, the raw data theoretically reaches the first half-arc of the land scan (e.g.,...). Figure 11 (as shown) and the second half-arc point cloud (as shown) Figure 12 As shown, the points should overlap and match each other. However, due to the laser scanning angle error of the airborne blue-green lidar depth sounding system, the actual visualized point cloud data has a certain positional deviation, which affects the subsequent underwater point cloud matching. Therefore, it is necessary to consider the system scanning angle deviation factor and correct the scanning angle of the rotating scanning system.

[0095] Furthermore, for the asynchronous observation point cloud obtained from the rotational scan, local scan point clouds containing obvious ground features are selected as target point clouds, such as locations with changes in road shape, building corners, and significant terrain changes. This involves extracting and filtering key feature points in the overlapping areas of the forward and backward scan semi-circular point clouds within the scanning flight strip, obtaining a certain number of pairs of identically named feature points as target point clouds. Table 1 below shows the asynchronous observation point clouds, where S represents the arc length and θ represents... The corresponding radians, Represents the central angle; after selecting and filtering feature point pairs, corresponding point pairs for asynchronous observation are obtained:

[0096] Table 1: Set of Key Point Extraction Pairs

[0097]

[0098] Furthermore, in another embodiment of the present invention, when probing a shallower body of water (2 meters deep), it is difficult to distinguish between the surface and bottom echoes through echo signal processing. Therefore, in this embodiment, the water depth is estimated using asynchronous observations within the flight path of an airborne blue-green lidar depth sounding system, and compared with the water depth values ​​obtained from simultaneous single-beam depth measurements. The comparison shows that refraction occurs at a water surface elevation of approximately 738.56 m, and the error between the laser point cloud and the water depth point is approximately 0.13 m, indicating that the laser detection accuracy meets expectations.

[0099] Furthermore, such as Figure 13 As shown, in another embodiment of the present invention, the statistical results of the comparison between the underwater topographic data of the unmanned vessel single beam and the corresponding position elevation deviation of the airborne blue-green lidar depth sounding system are presented. The mean is 0.08 and the standard deviation is 0.13. The specific data are shown in Table 2 below, which shows the statistical data of the water depth estimation deviation corresponding to the coordinate positions of some underwater checkpoints.

[0100] Table 2: Statistical Table of Deviations in Estimated Water Depth Corresponding to the Coordinates of Underwater Checkpoints

[0101]

[0102] This invention is based on the space detection principle and point cloud calculation method of airborne single-band depth sounding system, avoiding the error influence between multi-source heterogeneous data detection results such as multi-system integration, and effectively improving the ground detection accuracy and water-land integrated detection capability of airborne blue-green lidar depth sounding system.

[0103] Furthermore, such as Figure 14As shown, based on the above-mentioned shallow water depth estimation method based on asynchronous observation and scan angle correction, the present invention also provides a shallow water depth estimation system based on asynchronous observation and scan angle correction, wherein the shallow water depth estimation system based on asynchronous observation and scan angle correction includes:

[0104] The land bias estimation module 51 is used to perform multiple asynchronous observations of the land target area to obtain multiple pairs of corresponding points, and to construct an optimization objective function based on all the pairs of corresponding points;

[0105] The land scan optimization module 52 is used to iteratively optimize the results of all asynchronous observations according to the optimization objective function to obtain the corrected scan angle error;

[0106] The key point extraction module 53 is used to asynchronously observe the underwater target area to obtain the first underwater point cloud and the second underwater point cloud. The first underwater point cloud and the second underwater point cloud are divided into multiple super voxel regions, and multiple key point pairs are extracted to construct a sparse correspondence relationship.

[0107] The first key point filtering module 54 is used to construct the distance value of each key point pair according to the sparse correspondence, and filter all key point pairs according to all the distance values ​​to obtain multiple initial key point pairs with consistent distances.

[0108] The second key point filtering module 55 is used to construct a set of neighboring points of the first key point and the second key point in each initial key point pair, and to construct a difference value based on the two sets of neighboring points.

[0109] The water depth estimation module 56 is used to construct an optimization function using all the said difference values, and to determine the water depth value of the underwater target area based on the optimization function.

[0110] Furthermore, such as Figure 15 As shown, based on the above-mentioned method and system for estimating shallow water depth based on asynchronous observation and scanning angle correction, the present invention also provides a terminal, which includes a processor 10, a memory 20 and a display 30. Figure 15 Only some of the terminal components are shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0111] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory. In other embodiments, the memory 20 may be an external storage device of the terminal, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the terminal. Further, the memory 20 may include both internal and external storage units of the terminal. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code installed on the terminal. The memory 20 may also be used to temporarily store data that has been output or will be output. In one embodiment, the memory 20 stores a shallow water depth estimation program 40 based on asynchronous observation and scan angle correction. This shallow water depth estimation program 40 based on asynchronous observation and scan angle correction can be executed by the processor 10, thereby implementing the shallow water depth estimation method based on asynchronous observation and scan angle correction in this application.

[0112] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data, such as executing the shallow water depth estimation method based on asynchronous observation and scan angle correction.

[0113] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 30 is used to display information on the terminal and to display a visual user interface. The components of the terminal communicate with each other via a system bus.

[0114] In one embodiment, when the processor 10 executes the shallow water depth estimation program 40 based on asynchronous observation and scan angle correction in the memory 20, the steps of the shallow water depth estimation method based on asynchronous observation and scan angle correction described above are implemented.

[0115] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a shallow water depth estimation program based on asynchronous observation and scan angle correction, wherein when the shallow water depth estimation program based on asynchronous observation and scan angle correction is executed by a processor, it implements the steps of the shallow water depth estimation method based on asynchronous observation and scan angle correction as described above.

[0116] In summary, this invention provides a shallow water depth estimation method and related equipment based on asynchronous observation and scan angle correction. The method includes: performing multiple asynchronous observations on a land target area to obtain multiple pairs of corresponding points, and constructing an optimization objective function based on the pairs of corresponding points; iteratively optimizing the results of all asynchronous observations based on the optimization objective function to obtain the corrected scan angle error; performing asynchronous observations on an underwater target area to obtain a first underwater point cloud and a second underwater point cloud, dividing the first underwater point cloud and the second underwater point cloud into multiple supervoxel regions, and extracting multiple key point pairs to construct a sparse correspondence; constructing a distance value for each key point pair based on the sparse correspondence, filtering all key point pairs based on all distance values ​​to obtain multiple initial key point pairs with consistent distances; constructing neighboring point sets for the first and second key points in each initial key point pair, and constructing difference values ​​based on the two neighboring point sets; constructing an optimization function using all the difference values, and determining the water depth value of the underwater target area based on the optimization function. This invention is based on the space detection principle and point cloud calculation method of airborne single-band depth sounding system, avoiding the error influence between multi-source heterogeneous data detection results such as multi-system integration, and effectively improving the ground detection accuracy and water-land integrated detection capability of airborne blue-green laser radar depth sounding system in blue-green band.

[0117] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal that includes that element.

[0118] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.). The program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The computer-readable storage medium can be a memory, magnetic disk, optical disk, etc.

[0119] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for estimating water depth in shallow water areas based on asynchronous observation and scanning angle correction, characterized in that, The shallow water depth estimation method based on asynchronous observation and scan angle correction includes: Multiple asynchronous observations were performed on the land target area to obtain multiple pairs of corresponding points, and an optimization objective function was constructed based on all the pairs of corresponding points. The results of all asynchronous observations are iteratively optimized according to the aforementioned objective function to obtain the corrected scan angle error. Asynchronous observation of the underwater target area is performed to obtain the first underwater point cloud and the second underwater point cloud. The first underwater point cloud and the second underwater point cloud are divided into multiple super voxel regions, and multiple key point pairs are extracted to construct a sparse correspondence relationship. Based on the sparse correspondence, a distance value is constructed for each key point pair. All key point pairs are filtered based on all the distance values ​​to obtain multiple initial key point pairs with consistent distances. Construct neighboring point sets for the first and second key points in each initial key point pair, and construct difference values ​​based on the two neighboring point sets. An optimization function is constructed using all the aforementioned differences, and the water depth of the underwater target area is determined based on the optimization function.

2. The shallow water depth estimation method based on asynchronous observation and scanning angle correction according to claim 1, characterized in that, The process involves performing multiple asynchronous observations of the land target area to obtain multiple pairs of corresponding points, and constructing an optimization objective function based on all of these pairs of corresponding points. Specifically, this includes: Multiple asynchronous scans of the land target area were performed to obtain multiple first land observation point clouds and multiple second land observation point clouds; Based on each of the first land observation point cloud and each of the second land observation point clouds, construct multiple pairs of points with the same name; Based on all the pairs of points with the same name, construct the spatial positional bias of the asynchronous observations: ； in, This indicates the first scan point of asynchronous land observation. This indicates the second scan point of asynchronous land observation. Indicates the first laser emission point for asynchronous observation on land. This indicates the second laser emission point for asynchronous observation on land. Representing vector symbols, express arrive The vector, express arrive The vector, express arrive The vector, , and They represent x-coordinate, y-coordinate, and y-coordinate , and They represent x-coordinate, y-coordinate, and y-coordinate express arrive distance, express arrive distance, and These represent the first spatial incident angle and the second spatial incident angle of the asynchronous land observation, respectively. and These represent the first spatial azimuth and the second spatial azimuth of the asynchronous land observation, respectively. Indicates the scanning angle error; Using the aforementioned spatial position deviation, an optimization objective function is constructed: ； in, This indicates the number of pairs of nodes with the same name. Indicates to index, This represents the objective function to be optimized. This indicates taking the minimum value.

3. The shallow water depth estimation method based on asynchronous observation and scanning angle correction according to claim 2, characterized in that, The step of iteratively optimizing the results of all asynchronous observations based on the optimization objective function to obtain the corrected scan angle error specifically includes: The scanning angle error of each pair of corresponding points is defined as an individual, and multiple individuals are randomly selected as the initial population. The initial population is screened using the optimization objective function to obtain several excellent individuals whose scanning angle error is less than the preset scanning angle error; Perform crossover operations on all the aforementioned superior individuals to generate multiple offspring individuals, and perform mutation operations on all the aforementioned offspring individuals to generate an offspring population; The offspring population is iterated until the target offspring population is obtained, and the corrected scan angle error is selected from the target offspring population.

4. The shallow water depth estimation method based on asynchronous observation and scanning angle correction according to claim 1, characterized in that, The hypervoxel region includes a first hypervoxel region and a second hypervoxel region; The asynchronous observation of the underwater target area yields a first underwater point cloud and a second underwater point cloud. These two underwater point clouds are then divided into multiple supervoxel regions, and multiple key point pairs are extracted to construct a sparse correspondence. Specifically, this includes: Asynchronous observation of the underwater target area is performed to obtain a first underwater point cloud and a second underwater point cloud. The first underwater point cloud and the second underwater point cloud are then uniformly divided according to the method of minimizing the feature distance of the supervoxel to obtain multiple first supervoxel regions and multiple second supervoxel regions. Based on the feature distance, add corresponding feature labels to all first hypervoxel regions and all second hypervoxel regions; Determine the centroids in all the first and second super-voxel regions, and select multiple first keypoints corresponding to the first underwater point cloud and multiple second keypoints corresponding to the second underwater point cloud based on all the centroids: ； in, Represents the feature distance. Indicates the echo signal strength characteristic label, express The weight, Labels representing the difference in point cloud distances. express The weight, Indicates the feature label of the normal direction of the point cloud. express The weight, This indicates the distance between the centroid and the first or second critical point. In the second underwater point cloud, identify multiple second key points that are similar to the feature labels of the first key points in the multiple first supervoxel regions, and construct key point pairs based on the first key points and the second key points; Construct a sparse correspondence between the first keypoint and the second keypoint in each keypoint pair: ； in, Symbols representing similarity calculations. Feature description representing the first key point, The feature description representing the second key point, This represents the number of feature dimensions in a keypoint pair. The first key point dimensional features, The second key point dimensional features, Indicates to The index.

5. The shallow water depth estimation method based on asynchronous observation and scanning angle correction according to claim 1, characterized in that, The step of constructing distance values ​​for each keypoint pair based on the sparse correspondence, and filtering all keypoint pairs based on all distance values ​​to obtain multiple initial keypoint pairs with consistent distances, specifically includes: Based on the sparse correspondence, determine the first and second correspondences in each keypoint pair: ； ； in, Indicates the first correspondence. This indicates the second correspondence. Indicates feature dimension as The first key point, Indicates feature dimension as The second key point, Indicates feature dimension as The first key point, Indicates feature dimension as The second key point, This represents the index indicating the number of feature dimensions for the first keypoint. An index representing the number of feature dimensions for the second keypoint; Based on each of the first and second correspondences, calculate the distance value corresponding to the keypoint pair: ； in, This represents the distance value between keypoint pairs. Distance calculation symbol, This indicates taking the minimum value; The distance value of each keypoint pair is compared with a distance threshold. If the distance value exceeds the distance threshold, the keypoint pair is deleted, resulting in multiple initial keypoint pairs with consistent distances.

6. The shallow water depth estimation method based on asynchronous observation and scanning angle correction according to claim 5, characterized in that, The step of constructing neighboring point sets for the first and second keypoints in each initial keypoint pair, and constructing difference values ​​based on the two neighboring point sets, specifically includes: For each pair of initial keypoints, construct a corresponding set of neighboring points for the first keypoint and the second keypoint: ； in, The index representing the number of initial keypoint pairs. This indicates the number of initial keypoint pairs. Indicates the first The first or second keypoint in an initial keypoint pair; Construct a linear representation of all elements in each of the nearest neighbor sets: ； ； in, Indicates the first The weight coefficient of the first key point in an initial key point pair Indicates the first The first key point in an initial key point pair Indicates the first The weight coefficient of the second keypoint in the initial keypoint pair Indicates the first The second key point in an initial key point pair; Based on all the linear representations, construct the difference values ​​for the corresponding initial keypoint pairs: ； in, Indicates the difference value.

7. The shallow water depth estimation method based on asynchronous observation and scanning angle correction according to claim 6, characterized in that, The step of constructing an optimization function using all the aforementioned difference values, and determining the water depth of the underwater target area based on the optimization function, specifically includes: Among all the discrepancy values, the minimum discrepancy value is determined, and the initial keypoint pair corresponding to the minimum discrepancy value is defined as the target keypoint pair. The remaining initial keypoint pairs are then used to construct keypoint matching pairs. ； in, Indicates keypoint matching pairs, , and These represent the first keypoints of the first, second, and third initial keypoint pairs in the keypoint matching pair, respectively. , and These represent the second keypoints of the first, second, and third initial keypoint pairs in the keypoint matching pair, respectively. and These represent the first underwater point cloud and the second underwater point cloud, respectively. Based on the key point matching pairs, an optimization function for the underwater asynchronous observation bias of the target key point pairs is constructed: ； in, Represents the optimization function. express arrive The vector, This indicates the number of initial keypoint pairs in the keypoint matching pair. express index, This indicates the first key point of the target key point pair. Indicates the second key point of the target key point pair; Minimize the optimization function to estimate the water depth of the underwater target area: ； in, Indicates the water depth value. Indicates the distance threshold. This indicates the laser incident direction for underwater detection. This represents a constant, with a value of 1 or 2. This indicates the direction of the first laser incident point for underwater detection. This indicates the incident direction of the second laser for underwater detection. Indicates transpose. , and They represent x-coordinate, y-coordinate, and y-coordinate , and They represent x-coordinate, y-coordinate, and y-coordinate This represents the minimum propagation time of a laser signal in water. and These respectively represent laser signals in terms of and The propagation time in water at the time of incident.

8. A shallow water depth estimation system based on asynchronous observation and scan angle correction, characterized in that, The shallow water depth estimation system based on asynchronous observation and scan angle correction includes: The land bias estimation module is used to perform multiple asynchronous observations of the land target area to obtain multiple pairs of corresponding points, and to construct an optimization objective function based on all the pairs of corresponding points; The land scan optimization module is used to iteratively optimize the results of all asynchronous observations according to the optimization objective function to obtain the corrected scan angle error. The key point extraction module is used to asynchronously observe the underwater target area to obtain the first underwater point cloud and the second underwater point cloud. The first underwater point cloud and the second underwater point cloud are divided into multiple super voxel regions, and multiple key point pairs are extracted to construct a sparse correspondence relationship. The first key point filtering module is used to construct the distance value of each key point pair according to the sparse correspondence, and filter all key point pairs according to all the distance values ​​to obtain multiple initial key point pairs with consistent distances. The second key point filtering module is used to construct a set of neighboring points of the first key point and the second key point in each initial key point pair, and to construct a difference value based on the two sets of neighboring points. A depth estimation module is used to construct an optimization function using all the said difference values, and to determine the depth value of the underwater target area based on the optimization function.

9. A terminal, characterized in that, The terminal includes: a memory, a processor, and a shallow water depth estimation program based on asynchronous observation and scan angle correction stored in the memory and executable on the processor. When the shallow water depth estimation program based on asynchronous observation and scan angle correction is executed by the processor, it implements the steps of the shallow water depth estimation method based on asynchronous observation and scan angle correction as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a shallow water depth estimation program based on asynchronous observation and scan angle correction, which, when executed by a processor, implements the steps of the shallow water depth estimation method based on asynchronous observation and scan angle correction as described in any one of claims 1-7.

Images