A method and system for 3D geo-mapping of a side scan sonar

Abstract

The application provides a three-dimensional ground image forming method and system of a depth sounding side scan sonar, and the method comprises the following steps: corresponding the side scan image of the depth sounding side scan sonar with the corresponding sparse depth sounding data, converting the side scan image from the slant range to the ground range projection through the interpolation water depth geometric correction; carrying out gridding interpolation on the sparse depth sounding data, constructing the initial depth field and the gradient field with the same resolution as the side scan image as the starting condition of the iterative optimization; constructing the energy functional which fuses the consistency of light, the integrability and the sparse data fidelity term, alternately iteratively solving the gradient field and the height field to obtain the initial inversion topography with the texture information; and using the sparse depth sounding data to perform the global polynomial calibration and the frequency domain compensation correction on the initial inversion topography to obtain the final topography. The application has the advantages that: the method only uses the depth sounding side scan sonar, does not need to depend on the external independent depth sounding equipment, and effectively recovers the fine texture of the seafloor microrelief while maintaining the spatial position and water depth accuracy.

Description

Technical Field

[0001] This application belongs to the fields of marine acoustic seabed topography detection technology and seabed topography mapping technology, specifically relating to a three-dimensional topographic imaging method and system for bathymetric side-scan sonar. Background Technology

[0002] Bathymetric side-scan sonar (BSSS) systems can simultaneously acquire high-resolution seabed side-scan images and bathymetric data with a certain degree of accuracy, making them important equipment for seabed micro-topography detection and target identification. However, a significant contradiction exists in the data processing and application of BSSS systems: side-scan images use short-time energy averaging to obtain backscatter intensity information, resulting in extremely high vertical track resolution; while bathymetric data, for accurate estimation of direction of arrival (DOA), requires spatial smoothing using multiple snapshots or long sliding windows, such as... Figure 2 This results in a much lower vertical track resolution than side-scan images. Consequently, many small terrain features that are clearly visible in side-scan images cannot be effectively distinguished in depth sounding topographic maps.

[0003] To utilize high-resolution side-scan images for 3D terrain inversion, Shape from Shading (SFS) technology has become a research hotspot. It utilizes the brightness variations in an image to invert the relative height of a target surface, and is currently the mainstream research method. Horn et al. first proposed describing the SFS problem as an energy function using a minimization method, introducing smoothness constraints to solve the surface gradient, laying the foundation for this field. Addressing the limitation of this method in losing details in complex textures, Zheng et al. abandoned the traditional smoothness assumption and proposed an image gradient constraint method, effectively enhancing the recovery of edge features. To further solve the integral drift problem, Johnson and Hebert pioneered the use of sparse bathymetry data as external control points, adding bathymetry constraints after inversion. Building on this, Zhao Jianhu, Liu Hongbo, and others have recently proposed a fusion strategy that uses water depth measured by a single beam for frequency domain complementation after side-scan image inversion, combined with a linearized solution algorithm. However, existing methods mostly rely on external, independent bathymetry equipment, such as single-beam or multi-beam sonar, to provide constraint data. This not only increases the equipment cost of offshore operations but also introduces registration errors due to spatiotemporal inconsistencies caused by different sources, making it difficult to provide strict point-to-point constraints. Therefore, how to achieve high-resolution, high-precision three-dimensional seabed topographic imaging using the BSSS system's own data without relying on external bathymetry equipment is a critical problem that urgently needs to be solved. Summary of the Invention

[0004] The purpose of this application is to overcome the shortcomings of existing technologies that rely heavily on external, independent depth sounding equipment to provide constraint data.

[0005] To achieve the above objectives, this application proposes a three-dimensional terrain imaging method using bathymetric side-scan sonar, comprising: Step 1: Match the side-scan images of the bathymetry side-scan sonar with the sparse bathymetry data from the same source, and convert the side-scan images from slant range to ground distance projection through interpolation water depth geometric correction; Step 2: Perform gridded interpolation on the sparse depth sounding data to construct an initial depth field and its gradient field with the same resolution as the side scan image, as the starting condition for iterative optimization; Step 3: Construct an energy functional that integrates illumination consistency, integrability, and sparse data fidelity terms, and iteratively solve the gradient field and height field to obtain the initial inverted terrain with texture information; Step 4: Use sparse sounding data to perform global polynomial calibration and frequency domain compensation correction on the initial inverted terrain to obtain the final terrain.

[0006] As an improvement to the above method, step 2 includes: Grid interpolation is performed on sparse bathymetry data to generate an initial depth field with the same resolution as the side scan image. And calculate the initial gradient field. This serves as the starting condition for iterative optimization: ; in, Indicates grid coordinates The initial depth field at the location; ) indicates in grid coordinates The initial gradient at; Indicates the grid in Spatial step size in the direction; Indicates the grid in Spatial step size in the direction.

[0007] As an improvement to the above method, the energy functional in step 3, which integrates illumination consistency, integrability, and sparse data fidelity terms, is expressed as: ; ; ; ; in, Represents the energy functional; For illumination consistency; These are integrability constraint terms; For sparse data fidelity; The height field of the seabed surface; is the gradient vector of the height field; The true value provided for sparse bathymetry data; These are the integrability constraint weight coefficients; Weighting coefficients for data fidelity items; For sparse water depth masking functions; The integral interval of the image; respectively along Surface gradient in direction; pixel coordinates Normalized side-scan energy: ; ; in, For reflection model functions; The reflectivity of the seabed surface; The incident sound ray vector, T This is the transpose of a vector.

[0008] As an improvement to the above method, the alternating iterative solution of the gradient field and height field to obtain the initial inverted terrain with texture information includes: An operator splitting strategy is used for alternating iterative solution: Step 1: Fix the height field Optimize gradient field To minimize illumination error: The gradient descent method is used for updating, the first... The iteration is represented as: ; in, The learning rate; , For the (k+1)th iteration value; The second step is to fix the gradient field. Find the height field that best fits the gradient distribution. That is, solving the Poisson equation: ; Using the differential properties of the Fourier transform, solve the Poisson equation in the frequency domain: ; ; in, Fourier transform; For the (k+1)th iteration value; For height field The corresponding representation in the frequency domain; For corresponding spatial coordinates and Components in the frequency domain; The third step involves using a weighted fusion method to pull the reconstructed terrain towards the true depth benchmark. The calculation formula is as follows: ; in, Weighting to ensure data fidelity.

[0009] As an improvement to the above method, step 4 includes: A two-stage correction method is employed, which includes global polynomial calibration and frequency domain compensation. Using a sparse effective set of sounding points, find the optimal polynomial mapping function. ,in, Let be the order of the polynomial. These are the polynomial coefficients; Construct a least squares optimization problem and solve for the coefficients. The mapping is applied to the entire map to obtain the calibrated terrain. : ; ; Using a Gaussian low-pass filter Extracting the trend term from the residuals : ; in, Let be the standard deviation of the Gaussian kernel function. Defined as: ; Final terrain By overlaying trend items get: .

[0010] This application also provides a three-dimensional terrain imaging system based on the above method, the system comprising: The ground distance projection conversion module is used to correspond the side-scan images of the bathymetry side-scan sonar with the same source sparse bathymetry data, and convert the side-scan images from slant range to ground distance projection through interpolation water depth geometric correction. The module for obtaining iteration start conditions is used to perform gridded interpolation on sparse depth sounding data and construct an initial depth field and its gradient field with the same resolution as the side scan image, which serves as the starting condition for iterative optimization. The initial inverted terrain module is obtained to construct an energy functional that integrates illumination consistency, integrability and sparse data fidelity terms. The gradient field and height field are solved alternately and iteratively to obtain the initial inverted terrain with texture information. The final terrain module is used to perform global polynomial calibration and frequency domain compensation correction on the initial inverted terrain using sparse bathymetry data to obtain the final terrain.

[0011] Compared with existing technologies, the advantages of this application are: This invention utilizes high-resolution side-scan images in the vertical track direction for SFS inversion and employs homologous sparse depth values ​​for constraint. This method does not rely on external independent bathymetry equipment, effectively recovering the fine texture of seabed micro-topography and improving spatial location and depth accuracy. It provides a novel method for obtaining high-precision, high-resolution three-dimensional topography in the vertical track direction for seabed topography exploration. Attached Figure Description

[0012] Figure 1 The diagram shown is a flowchart of the invention method; Figure 2 The diagram shown is a schematic of the side scan and depth sounding processing window; Figure 3(a) shows the region-1 inversion result - the original side scan image; Figure 3(b) shows the inversion results for region 1 - sparse sounding points; Figure 3(c) shows the inversion results for region 1 and the final inversion results; Figure 3(d) shows a three-dimensional plot of the region-1 inversion result and the final inversion result; Figure 3(e) shows the results of the region-one inversion and the sparse sounding interpolation results; Figure 3(f) shows a three-dimensional plot of the region-inversion results and sparse sounding interpolation results; Figure 4(a) shows the results of the region 2 inversion - the original side scan image; Figure 4(b) shows the results of the two-way inversion of the region - sparse sounding points; Figure 4(c) shows the inversion results for Region 2 - the final inversion results; Figure 4(d) shows a three-dimensional plot of the inversion results in Region 2 and the final inversion results; Figure 4(e) shows the results of the two-way inversion of the region - the results of sparse sounding interpolation; Figure 4(f) shows a three-dimensional plot of the sparse sounding interpolation results of the region 2 inversion; Figure 5(a) shows the starboard reprojection image of the inversion results of the two geometric correction methods - the original side scan image; Figure 5(b) shows the starboard reprojection of the inversion results of the two geometric correction methods - geometric correction for fixed water depth per ping; Figure 5(c) shows the starboard reprojection of the inversion results of two geometric correction methods - geometric correction of the method of the present invention; Figure 6(a) shows the port side reprojection image of the inversion results of the two geometric correction methods - the original side scan image; Figure 6(b) shows the port side reprojection of the inversion results of the two geometric correction methods - geometric correction for fixed water depth per ping; Figure 6(c) shows the port side reprojection diagram of the inversion results of two geometric correction methods - geometric correction of the method of the present invention; Figure 7(a) shows the inversion depth error distribution - the coincidence error distribution within the results of region one; Figure 7(b) shows the inversion depth error distribution - the external coincidence error distribution of the results in region one; Figure 7(c) shows the inversion depth error distribution - the coincidence error distribution within the results of Region 2; Figure 7(d) shows the inversion depth error distribution - the external coincidence error distribution of the results in Region 2. Detailed Implementation

[0013] The technical solution of this application will be described in detail below with reference to the accompanying drawings.

[0014] This application provides a three-dimensional terrain imaging method and system based on a bathysound side-scan sonar, which is implemented based on the same source sparse bathysound constraint. The method mainly includes: ① mapping the high-resolution side-scan image of the bathysound side-scan sonar to the same source sparse bathysound data, and converting the side-scan image from slant range to ground distance projection through interpolation water depth geometric correction; ② performing gridded interpolation on the sparse bathysound data to construct an initial depth field and its gradient field with the same resolution as the side-scan image, as the starting condition for subsequent iterative optimization; ③ constructing an energy functional that integrates illumination consistency, integrability and sparse data fidelity terms, and alternately iteratively solving the gradient field and height field to obtain the initial inverted terrain with texture information; ④ using the sparse bathysound data to perform global polynomial calibration and frequency domain compensation correction on the initial inverted terrain to obtain the final terrain.

[0015] Example 1 like Figure 1 As shown, the three-dimensional terrain imaging method of a depth-sounding side-scan sonar provided in this application includes: 1. Preprocessing side-scan images and sparse depth data High-resolution side-scan images are generated based on the echo signals from the bathysound side-scan sonar. Sparse depth values ​​are obtained through DOA estimation and then mapped accordingly. The side-scan images are segmented using a thresholding method to extract the seabed line and remove blind spots in the water body. Traditional side-scan sonar SFS inversion methods often lack synchronous bathysound data support, therefore using the seabed line to calculate the depth as a fixed depth per ping for geometric correction, which results in geometric distortion. This invention's method is based on a bathysound side-scan sonar system, thus utilizing depth values ​​from the same source interpolation to complete the geometric correction of the side-scan image from slant range to horizontal range. A local Cartesian coordinate system is established with the initial sonar position as the origin, where... The shaft is along the direction of the carrier's travel. The axis is perpendicular to the heading and points to starboard. The axis is vertically upward.

[0016] (1) in, For the first i 1 pixel horizontal coordinate, For the first i Slope distance of 1 pixel, For the first i The water depth values ​​after sparse water depth interpolation of each pixel. The effects of the two geometric correction methods on the final results are shown in Figures 5(a)-5(c) and 6(a)-6(c).

[0017] 2. Calculate the initial depth field and its gradient field at the same resolution as the side scan image. Grid interpolation is performed on sparse water depth data to generate an initial depth field with the same resolution as the side scan image. And calculate the initial gradient field. This serves as the starting condition for iterative optimization.

[0018] (2) in, Indicates grid coordinates The initial depth field at the location; Indicates grid coordinates The initial gradient at; Indicates the grid in Spatial step size in the direction; Indicates the grid in Spatial step size in the direction.

[0019] 3. Initial terrain inversion based on the SFS method For rough surfaces, the scattering mechanism of sound waves follows the Lambertian law of reflection. Assuming the seabed surface satisfies the Lambertian reflection model, then: (3) (4) in, pixel coordinates Normalized side-scan energy; For reflection model functions; The reflectivity of the seabed surface is usually assumed to be constant; The angle between the incident sound ray vector and the normal vector of the seabed surface is; the incident sound ray vector is The height field of the seabed surface is The normal vector of the seabed surface is , respectively along Surface gradient in the direction.

[0020] To recover high-resolution 3D seabed topography from side-scan images, we transform the reconstruction problem in Equation (3) into an energy minimization problem, introducing sparse bathymetry data as a constraint on top of SFS. The following is constructed, including an illumination consistency term. Integrability constraint terms and sparse data fidelity items The energy functional consists of three parts: (5) (6) (7) (8) in, is the gradient vector of the height field; The true value provided for sparse bathymetry data; These are the integrability constraint weight coefficients; Weighting coefficients for data fidelity items; For sparse water depth masking functions, Let be the integral interval of the image. The gradient field obtained from the constraint inversion must conform to the physical laws of sonar imaging. Ensure the existence of a continuously differentiable height surface. The inversion results are constrained by sparse depth values ​​to prevent divergence. In equation (3) This describes the Lambertian scattering model under ideal, noise-free conditions. However, real-world 3D terrain reconstruction problems involve noise, and... Since the unknown is the one that needs to be solved, there is an error between the two. Therefore, we construct equation (6) to find the minimum value of this error, thereby determining the value of the unknown. .

[0021] Due to the above functional The problem is non-convex and highly nonlinear, making direct simultaneous solution extremely difficult. This invention employs an operator splitting strategy, decomposing the original problem into three sub-problems for alternating iterative solutions. The first step involves fixing the height field. Optimize gradient field To minimize illumination error. Due to the reflection function It exhibits nonlinearity and is updated using the gradient descent method. According to the differential relationship in equation (9), the... The next iteration is as shown in equation (10). is the learning rate. To ensure numerical stability, a small local mean smoothing operation is introduced in this step to suppress noise.

[0022] (9) (10) The second step is to fix the gradient field. Find the height field that best fits the gradient distribution. This is equivalent to solving the Poisson equation (11). To avoid error accumulation caused by difference iteration and to improve computational efficiency, the equation is solved in the frequency domain using the differential properties of the Fourier transform: (11) (12) (13) in, Fourier transform; For height field The corresponding representation in the frequency domain; Corresponding spatial coordinates and Components in the frequency domain.

[0023] The third step is to use a weighted fusion method to pull the reconstructed terrain towards the true depth benchmark. The calculation formula is as follows: (14) in, Weighting to ensure data fidelity.

[0024] 4. Correction under sparse water depth constraints although While reconstructing detailed micro-topographic features, the absolute depth and macroscopic trends of SFS (Sparse Sounding System) may still deviate from the actual terrain due to inherent scale ambiguity. Therefore, this invention employs a two-stage correction method involving global polynomial calibration and frequency domain compensation. First, the optimal polynomial mapping function is found using a sparse set of effective sounding points. This is to correct the overall nonlinear distortion of the SFS results. Let be the order of the polynomial. The coefficients are polynomial coefficients. The following least squares optimization problem is constructed as shown in equation (15), and the coefficients are solved... Applying this mapping to the entire map yields the calibrated terrain. : (15) (16) Since the SFS algorithm is required to recover terrain details, i.e., high-frequency terrain information, it is assumed that the SFS inversion result... With sparse truth value The residuals are mainly composed of low-frequency errors, which can be addressed using a Gaussian low-pass filter. Extracting the trend term from the residuals: (17) in, It is the standard deviation of the Gaussian kernel function. Defined as: (18) Final high-precision terrain This is achieved by overlaying the trend term, thus effectively combining macroscopic precision with microscopic texture: (19) Figures 3(a)-3(f) and 4(a)-4(f) show comparisons between the inversion results and direct sparse interpolation results for the seabed of Region 1 and Region 2 using the method of this invention. Figures 7(a)-7(d) show histograms of the inversion depth error distribution. Table 1 shows the inversion depth accuracy results.

[0025] Table 1. Accuracy of Inverted Water Depth

[0026] Example 2 This application also provides a three-dimensional terrain imaging system based on the above method, the system comprising: The ground distance projection conversion module is used to correspond the side-scan images of the bathymetry side-scan sonar with the same source sparse bathymetry data, and convert the side-scan images from slant range to ground distance projection through interpolation water depth geometric correction. The module for obtaining iteration start conditions is used to perform gridded interpolation on sparse depth sounding data and construct an initial depth field and its gradient field with the same resolution as the side scan image, which serves as the starting condition for iterative optimization. The initial inverted terrain module is obtained to construct an energy functional that integrates illumination consistency, integrability and sparse data fidelity terms. The gradient field and height field are solved alternately and iteratively to obtain the initial inverted terrain with texture information. The final terrain module is used to perform global polynomial calibration and frequency domain compensation correction on the initial inverted terrain using sparse bathymetry data to obtain the final terrain.

[0027] This application may also provide a computer device, including: at least one processor, memory, at least one network interface, and a user interface. The various components in this device are coupled together via a bus system. It is understood that the bus system is used to implement communication between these components. In addition to a data bus, the bus system also includes a power bus, a control bus, and a status signal bus.

[0028] The user interface can include a display, keyboard, or clicking device. Examples include a mouse, trackball, touchpad, or touchscreen.

[0029] It is understood that the memory in the embodiments disclosed in this application may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memories described herein are intended to include, but are not limited to, these and any other suitable types of memory.

[0030] In some implementations, the memory stores elements such as executable modules or data structures, or subsets thereof, or extended sets thereof: operating systems and applications.

[0031] The operating system includes various system programs, such as the framework layer, core library layer, and driver layer, used to implement various basic business functions and handle hardware-based tasks. The application programs include various applications, such as media players and browsers, used to implement various application functions. Programs implementing the methods of the embodiments of this disclosure can be included in the application programs.

[0032] In the above embodiments, the processor can also invoke programs or instructions stored in memory, specifically programs or instructions stored in an application program, for the following purposes: Follow the steps described above.

[0033] The above methods can be applied to or implemented by a processor. The processor may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above methods can be completed by integrated logic circuits in the processor's hardware or by software instructions. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic diagrams disclosed above. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the disclosed methods can be directly implemented by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory; the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above methods.

[0034] It is understood that the embodiments described in this application can be implemented using hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions described in this application, or combinations thereof.

[0035] For software implementation, the technology of this application can be implemented by executing the functional modules (e.g., procedures, functions, etc.) of this application. The software code can be stored in memory and executed by a processor. The memory can be implemented in the processor or outside the processor.

[0036] This application may also provide a non-volatile storage medium for storing a computer program. When the computer program is executed by a processor, it can implement the steps in the above method embodiments.

[0037] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application and are not intended to limit it. Although this application has been described in detail with reference to the embodiments, those skilled in the art should understand that modifications or equivalent substitutions to the technical solutions of this application do not depart from the spirit and scope of the technical solutions of this application, and should all be covered within the scope of the claims of this application.

Claims

1. A three-dimensional terrain imaging method using a bathysounding side-scan sonar, comprising: Step 1: Match the side-scan images of the bathymetry side-scan sonar with the sparse bathymetry data from the same source, and convert the side-scan images from slant range to ground distance projection through interpolation water depth geometric correction; Step 2: Perform gridded interpolation on the sparse depth sounding data to construct an initial depth field and its gradient field with the same resolution as the side scan image, as the starting condition for iterative optimization; Step 3: Construct an energy functional that integrates illumination consistency, integrability, and sparse data fidelity terms, and iteratively solve the gradient field and height field to obtain the initial inverted terrain with texture information; Step 4: Use sparse bathymetry data to perform global polynomial calibration and frequency domain compensation correction on the initial inverted terrain to obtain the final terrain; In step 3, the energy functional that integrates illumination consistency, integrability, and sparse data fidelity terms is constructed and expressed as: ； ； ； ； in, Represents the energy functional; For illumination consistency; These are integrability constraint terms; For sparse data fidelity; The height field of the seabed surface; is the gradient vector of the height field; The true value provided for sparse bathymetry data; These are the integrability constraint weight coefficients; Weighting coefficients for data fidelity items; For sparse water depth masking functions; The integral interval of the image; respectively along Surface gradient in direction; pixel coordinates Normalized side-scan energy: ； ； in, For reflection model functions; The reflectivity of the seabed surface; The incident sound ray vector, T Transpose of a vector; The alternating iterative solution of the gradient field and height field yields an initial inverted terrain with texture information, including: An operator splitting strategy is used for alternating iterative solution: Step 1: Fix the height field Optimize gradient field To minimize illumination error: The gradient descent method is used for updating, the first... The iteration is represented as: ； in, The learning rate; , For the (k+1)th iteration value; The second step is to fix the gradient field. Find the height field that best fits the gradient distribution. That is, solving the Poisson equation: ； Using the differential properties of the Fourier transform, solve the Poisson equation in the frequency domain: ； ； in, Fourier transform; For the (k+1)th iteration value; For height field The corresponding representation in the frequency domain; For corresponding spatial coordinates and Components in the frequency domain; The third step involves using a weighted fusion method to pull the reconstructed terrain towards the true depth benchmark. The calculation formula is as follows: ； in, Weighting to ensure data fidelity.

2. The three-dimensional terrain imaging method of the bathymetric side-scan sonar according to claim 1, characterized in that, Step 2 includes: Grid interpolation is performed on sparse bathymetry data to generate an initial depth field with the same resolution as the side scan image. And calculate the initial gradient field. This serves as the starting condition for iterative optimization: ； in, Indicates grid coordinates The initial depth field at the location; ) indicates in grid coordinates The initial gradient at; Indicates the grid in Spatial step size in the direction; Indicates the grid in Spatial step size in the direction.

3. The three-dimensional terrain imaging method of the bathymetric side-scan sonar according to claim 1, characterized in that, Step 4 includes: A two-stage correction method is employed, which includes global polynomial calibration and frequency domain compensation. Using a sparse effective set of sounding points, find the optimal polynomial mapping function. ,in, Let be the order of the polynomial. These are the polynomial coefficients; Construct a least squares optimization problem and solve for the coefficients. The mapping is applied to the entire map to obtain the calibrated terrain. : ； ； Using a Gaussian low-pass filter Extracting the trend term from the residuals : ； in, Let be the standard deviation of the Gaussian kernel function. Defined as: ； Final terrain By overlaying trend items get: 。 4. A three-dimensional terrain imaging system based on a depth-sounding side-scan sonar, implemented according to any one of claims 1-3, characterized in that, The system includes: The ground distance projection conversion module is used to correspond the side-scan images of the bathymetry side-scan sonar with the same source sparse bathymetry data, and convert the side-scan images from slant range to ground distance projection through interpolation water depth geometric correction. The module for obtaining iteration start conditions is used to perform gridded interpolation on sparse depth sounding data and construct an initial depth field and its gradient field with the same resolution as the side scan image, which serves as the starting condition for iterative optimization. The initial inverted terrain module is obtained, which is used to construct an energy functional that integrates illumination consistency, integrability, and sparse data fidelity terms. The gradient field and height field are then iteratively solved alternately to obtain the initial inverted terrain with texture information. The final terrain module is used to perform global polynomial calibration and frequency domain compensation correction on the initial inverted terrain using sparse bathymetry data to obtain the final terrain.

Images