Side-scan sonar image three-dimensional reconstruction method

By designing an adaptive hybrid reflection model and inversion elevation constraint coefficients, the problem of low three-dimensional reconstruction accuracy of side-swept sonar images is solved, and the submarine topographic information recovery is achieved with higher accuracy and efficiency, which is suitable for complex submarine environments.

CN118962688APending Publication Date: 2024-11-15JIANGSU UNIV OF SCI & TECH
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202411083322.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-08
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The existing three-dimensional reconstruction method of side-sweep sonar images cannot effectively restore the three-dimensional information of the seabed terrain, especially in complex environments, which are difficult to improve accuracy and accuracy.

Method used

A three-dimensional reconstruction method for side-swept sonar images is designed, including designing an adaptive hybrid reflection model, using the adaptive specular reflection coefficient, diffuse reflection coefficient and reflection index, combined with the adaptive hybrid reflection model expression of the seabed surface, and adopting the inversion elevation. Constraint coefficients to improve the accuracy of three-dimensional reconstruction.

Benefits of technology

By improving the matching of the side-sweep sonar image and the reflection model of the seabed surface, the dependence of three-dimensional reconstruction on external data is reduced, the accuracy and efficiency of three-dimensional reconstruction in complex seabed environments is improved, and more intuitive and reliable underwater three-dimensional terrain information is provided. .

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118962688A_ABST
    Figure CN118962688A_ABST
Patent Text Reader

Abstract

The invention discloses a side-scan sonar image three-dimensional reconstruction method. The method comprises the following steps: firstly, reading a side-scan sonar image to be subjected to three-dimensional reconstruction; secondly, designing a side-scan sonar image SFS mixed reflection model; comprising the following steps: designing a self-adaptive mirror reflection coefficient Dg; designing a self-adaptive diffuse reflection coefficient Dr; designing a self-adaptive specular reflection index epsilon and designing a seabed surface self-adaptive mixed reflection model expression; then calculating a three-dimensional reconstruction surface elevation; comprising the steps of inverting a seabed surface elevation value and designing an inversion elevation constraint coefficient ki; and finally, outputting a side-scan sonar image three-dimensional reconstruction result. Through the above steps, the side-scan sonar image three-dimensional reconstruction method improves the submarine topography three-dimensional reconstruction precision, and provides underwater three-dimensional topography information with higher quality for AUV underwater accurate navigation, underwater rescue, submarine exploration and other application scenes.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of sonar image processing, and in particular to a three-dimensional reconstruction method for side-scan sonar images. Background Art

[0002] Side-scan sonar is widely used in seabed exploration, underwater rescue, and marine surveys because it can obtain high-resolution seabed topographic images economically and efficiently. However, due to the limitations of the imaging mechanism, side-scan sonar cannot directly obtain three-dimensional information of the seabed topography, and it is difficult to directly express the ups and downs of the seabed topography. It is necessary to restore the three-dimensional information of the seabed topography through three-dimensional reconstruction. Therefore, how to reconstruct the three-dimensional information of side-scan sonar images has become a research hotspot in the field of hydroacoustics.

[0003] Existing three-dimensional reconstruction of side-scan sonar images can be divided into two methods, namely the shadow length recovery method and the shape from shading (SFS) method. The former is limited to the recovery of a single target and cannot carry out large-scale three-dimensional reconstruction of terrain; the latter mainly uses the light and shade changes of objects in the side-scan image to restore the three-dimensional terrain of the seabed, which is suitable for three-dimensional reconstruction of the seabed terrain in large areas. Therefore, three-dimensional reconstruction research based on the SFS method has been carried out. In the SFS method, the reflection model is the focus of three-dimensional reconstruction. How to make the reflection model more consistent with the characteristics of the seabed surface has become the key to three-dimensional reconstruction of terrain. Therefore, the research on the three-dimensional reconstruction method of side-scan sonar images is helpful to restore the stereo information of side-scan sonar images with high precision and improve the accuracy of three-dimensional reconstruction of side-scan sonar images in complex seabed environments. Summary of the invention

[0004] Purpose of the invention: To provide a method for three-dimensional reconstruction of side-scan sonar images, so as to improve the quality of stereo information restoration of side-scan sonar images, and thereby improve the accuracy of three-dimensional reconstruction of side-scan sonar images.

[0005] Technical solution: A method for three-dimensional reconstruction of side-scan sonar images, comprising the following steps:

[0006] Step (1): reading the side scan sonar image to be three-dimensionally reconstructed;

[0007] Step (2): Design the SFS mixed reflection model of the side scan sonar image;

[0008] The side scan sonar image SFS mixed reflection model is designed; including designing an adaptive mirror reflection coefficient D g ; Design adaptive diffuse reflectance D r ; Design an adaptive specular reflection index ε; Design an adaptive mixed reflection model expression for the seafloor surface;

[0009] Step (3): Obtain the elevation of the three-dimensional reconstructed surface;

[0010] The method of obtaining the three-dimensional reconstructed surface elevation includes inverting the seabed surface elevation value and designing the inverted elevation constraint coefficient k. i ;

[0011] Step (4): Output the three-dimensional reconstruction result of the side-scan sonar image.

[0012] Furthermore, in step (2), the design of the adaptive mirror reflection coefficient D g , specifically including the following steps:

[0013] 1) Calculate the surface roughness C of the side scan sonar image;

[0014] 2) Adaptive mirror reflection coefficient D g , using the following formula:

[0015]

[0016] Furthermore, in step (2), the design of the adaptive diffuse reflectance coefficient D r , using the following formula:

[0017]

[0018] Furthermore, in step (2), the design of the adaptive specular reflection index ε is specifically based on the following formula:

[0019]

[0020] Where: λ is a constant.

[0021] Furthermore, in step (2), the expression of the seabed surface adaptive mixed reflection model is designed, and the specific formula is as follows:

[0022]

[0023] Where: R hs (p,q) is the adaptive mixed reflection model function of the seabed surface, p and q are the horizontal and vertical gradients of the seabed surface, respectively. τ is the inclination and deflection of the incident direction of the sonar signal, ρ is the intensity of the incident sound wave, and p s ,q s are the horizontal and vertical gradients of the energy source position, respectively.

[0024] Furthermore, in designing the expression of the seabed surface adaptive mixed reflection model, the following steps are used to determine the parameters: τ:

[0025] a) Set the y-axis of the coordinate system of a point on the seabed surface to be in the same direction as the side-scan sonar track line, and the incident deflection angle τ of the side-scan sonar signal is 0;

[0026] b) According to the working mechanism of the side scan sonar, the corresponding inclination angle of the pixel at the xth position in the i-th frame of the side scan sonar image can be obtained: for:

[0027]

[0028] Where: arctan(·) is the inverse tangent function; W i is the position of the seabed point in the i-th frame of the side scan sonar image;

[0029] c) Calculate the inclination angle of all pixels in each frame of the side-scan sonar image, and then obtain the corresponding inclination angle of the pixels of the entire side-scan sonar image.

[0030] Furthermore, in step (3), the inversion of the seafloor surface elevation value specifically adopts the following steps:

[0031] (A) The backward finite difference method is used to discretize the seafloor surface gradients p and q approximately as follows:

[0032]

[0033] Where: Z(x,y) is the elevation of the seafloor surface at position (x,y); p and q are the horizontal and vertical gradients of the seafloor surface, respectively;

[0034] (B) The radiation illumination equation of the seabed surface in the side scan sonar image can be written as:

[0035] f[Z(x,y)]=0≈L(x,y)-R(p(x,y),q(x,y))

[0036] Where R(·) represents the reflection model function;

[0037] (C) Taylor expansion of a pixel point (x, y) on the side scan sonar image yields:

[0038]

[0039] (D) Let Z(x,y)=Z n (x,y), then the surface elevation value of the nth iteration is:

[0040]

[0041] Where: Z n (x,y),Z n-1 (x, y) are the surface elevation iteration results of the nth and n-1th times respectively;

[0042] (E) Setting the seafloor surface elevation iteration termination condition μ, that is, when the following expression is satisfied, the inversion of the seafloor surface elevation value is completed;

[0043] Z n (x,y)-Z n-1 (x,y)≤μ

[0044] Where: μ is the iteration termination condition of the seabed surface elevation.

[0045] Furthermore, in step (3), the design inversion elevation constraint coefficient k i , specifically the following steps:

[0046] (I) Set the seabed point position of the i-th frame of the side scan sonar image to W i , then the current depth of the AUV from the seabed is H i for:

[0047]

[0048] Where: U is the single-side scan width of the side scan sonar; N tra is the lateral pixel distance of the side scan image;

[0049] (II) Design the following inversion elevation constraint coefficient k for the i-th frame of the side scan sonar image: i :

[0050]

[0051] Where: It is the inverted elevation value of the track line position of the i-th frame of the sidescan sonar image.

[0052] Furthermore, in step (3), the three-dimensional reconstructed surface elevation is obtained by specifically adopting the following steps:

[0053] (a) Invert the seabed surface elevation of the side scan sonar image;

[0054] (b) Find all the depth data on the track line, then the depth set H = {H1, H2…, H i …};

[0055] (c) Calculate the inversion elevation constraint coefficient k of the i-th frame of the side scan sonar image i ;

[0056] (d) The expression of the inverted elevation value Φ(Z(x,i)) corresponding to the i-th frame of the constrained side scan sonar image is as follows:

[0057]

[0058] Where: Z(x,i) is the inverted elevation value corresponding to the i-th frame of the side scan sonar image;

[0059] (e) After all frames of the i-th side-scan sonar image have been constrained in their corresponding elevation values, the elevation values ​​of other frames of the side-scan sonar image are constrained in turn, thereby completing the elevation scale constraint of the seabed surface of the side-scan sonar image.

[0060] Beneficial effects: Compared with the prior art, the present invention first reads the side scan sonar image to be three-dimensionally reconstructed; secondly, designs the SFS mixed reflection model of the side scan sonar image; including designing the adaptive mirror reflection coefficient D g ; Design adaptive diffuse reflectance D r ; Design the adaptive specular reflection index ε and the adaptive mixed reflection model expression of the seafloor surface; then obtain the three-dimensional reconstructed surface elevation; including inverting the seafloor surface elevation value and designing the inversion elevation constraint coefficient k i ; Finally, the 3D reconstruction result of the side-scan sonar image is output; Through the above steps, the 3D reconstruction method of the side-scan sonar image of the present invention helps to restore the stereo information of the side-scan sonar image with high precision, improves the 3D reconstruction accuracy of the side-scan sonar image in a complex seabed environment, provides more intuitive and reliable underwater 3D terrain information, and provides higher quality underwater 3D terrain information for application scenarios such as AUV underwater precise navigation, underwater rescue, and seabed exploration, specifically including the following advantages:

[0061] (1) The SFS mixed reflection model of side-scan sonar images designed by the present invention is based on the characteristics of the continental shelf seabed, which is relatively flat and has reefs, and includes both diffuse reflection and specular reflection. The present invention retains the Lambert diffuse reflection model and introduces the Blinn-Phong specular reflection model, forming an SFS mixed reflection model that is adaptive to complex seabed surfaces, enhancing the matching of the reflection model with the seabed surface, thereby improving the accuracy of three-dimensional reconstruction of side-scan sonar images.

[0062] (2) The adaptive mixed reflection model expression of the seabed surface designed by the present invention adopts the exponential function form to design the adaptive mirror reflection coefficient D g , using the fraction containing the roughness C of the side scan sonar image to express D g The base number is set, and the roughness C is used as an index to improve the correlation between the side scan sonar image and the seabed surface reflection model; an adaptive specular reflection index ε is designed to convert D g As the formula coefficient of the specular reflection index ε, the inverse of the roughness is multiplied by the constant λ to express ε, and the adaptive diffuse reflection coefficient D is designed r , which improves the adaptability of the SFS mixed reflection model and thus improves the accuracy of three-dimensional reconstruction of side-scan sonar images.

[0063] (3) The inversion elevation constraint coefficient k designed by the present invention i, the depth data set H = {H1, H2…, H i …,H o}, and use the depth data and track line position to invert the elevation value To design the inversion height constraint coefficient k i , with this coefficient k i The inverted elevation value Z(x,i) is constrained to obtain the absolute elevation value Φ(Z(x,i)), which reduces the dependence of the three-dimensional reconstruction of side-scan sonar images on external side-scan data and enhances the role of the mixed reflection model in three-dimensional reconstruction, thereby improving the efficiency and accuracy of three-dimensional reconstruction of side-scan sonar images. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A general flow chart of a method for three-dimensional reconstruction of side-scan sonar images of the present invention;

[0065] Figure 2 Map of seafloor surface distribution characteristics;

[0066] Figure 3 Comparison of side scan sonar images I and 3D reconstruction. Figure 3 (a) in the figure shows the side scan sonar image I. Figure 3 (b) shows the 3D reconstruction using the Lambert model method. Figure 3 (c) in the figure shows the 3D reconstruction using the Ward hybrid model method. Figure 3 (d) shows the three-dimensional reconstruction diagram of the method of the present invention;

[0067] Figure 4 Comparison of side scan sonar image II 3D reconstruction, Figure 4 (a) in the figure shows the side scan sonar image II. Figure 4 (b) shows the 3D reconstruction using the Lambert model method. Figure 4 (c) in the figure shows the 3D reconstruction using the Ward hybrid model method. Figure 4 (d) shows the three-dimensional reconstruction diagram of the method of the present invention;

[0068] Figure 5 Comparison of side scan sonar images III 3D reconstruction, Figure 5 (a) in the figure shows the side scan sonar image III. Figure 5 (b) shows the 3D reconstruction using the Lambert model method. Figure 5 (c) in the figure shows the 3D reconstruction using the Ward hybrid model method. Figure 5 (d) shows the three-dimensional reconstruction diagram of the method of the present invention;

[0069] Figure 6 Side scan sonar image IV 3D reconstruction comparison chart, Figure 6(a) in the figure shows the side scan sonar image IV. Figure 6 (b) shows the 3D reconstruction using the Lambert model method. Figure 6 (c) in the figure shows the 3D reconstruction using the Ward hybrid model method. Figure 6 (d) in the figure shows the three-dimensional reconstruction image of the method of the present invention. DETAILED DESCRIPTION

[0070] In order to make the purpose and technical solution of the embodiments of the present invention clearer, the technical solution of the present invention is further described in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] like Figure 1 As shown, a method for 3D reconstruction of side-scan sonar images of the present invention is used to improve the quality of stereo information restoration of side-scan sonar images, thereby improving the accuracy of 3D reconstruction of side-scan sonar images. First, the side-scan sonar image to be 3D reconstructed is read; secondly, the SFS mixed reflection model of the side-scan sonar image is designed; then the elevation of the 3D reconstruction surface is obtained; and finally, the 3D reconstruction result of the side-scan sonar image is output. The specific steps are as follows:

[0072] Step 1. Read the side scan sonar image to be 3D reconstructed.

[0073] Step 2. Design the SFS mixed reflection model of side scan sonar images.

[0074] The side scan sonar image SFS mixed reflection model is designed; including designing an adaptive mirror reflection coefficient D g ; Design adaptive diffuse reflectance D r ; Design an adaptive mirror reflection index ε; Design an adaptive mixed reflection model expression for the seabed surface. Side-scan sonar can obtain high-resolution two-dimensional images of the seabed, but it cannot intuitively express the seabed terrain information. Relevant studies have shown that the echo intensity of the side-scan sonar is related to the undulations of the seabed terrain, and can reflect information related to the seabed terrain. If more seabed terrain information can be mined based on this, it will provide a basis for the reconstruction of the seabed terrain. In order to make up for the defect that side-scan sonar images cannot intuitively express the seabed terrain, the side-scan sonar images are reconstructed in three dimensions through the shading recovery shape method (SFS). Figure 2 As shown, the seabed surface of the continental shelf is relatively flat and has reefs, etc., which includes both diffuse reflection and specular reflection. Therefore, the present invention introduces the Blinn-Phong model to represent the specular reflection of the seabed while retaining the Lambert model to represent the diffuse reflection of the seabed surface.

[0075] The designed adaptive specular reflection coefficient Dg , specifically the following steps:

[0076] 1) Calculate the surface roughness C of the side scan sonar image.

[0077] 2) Adaptive mirror reflection coefficient D g , using the following formula:

[0078]

[0079] The design of the adaptive specular reflection index ε specifically adopts the following formula:

[0080]

[0081] Where: λ is a constant.

[0082] The designed adaptive diffuse reflectance D r , using the following formula:

[0083]

[0084] The design of the seabed surface adaptive mixed reflection model expression specifically adopts the following steps:

[0085] a) Design the following seabed surface adaptive mixed reflection model expression.

[0086]

[0087] Where: R hs (p,q) is the adaptive mixed reflection model function of the seabed surface, p and q are the horizontal and vertical gradients of the seabed surface, respectively. τ is the inclination and deflection of the incident direction of the sonar signal, ρ is the intensity of the incident sound wave, and p s ,q s are the horizontal and vertical gradients of the energy source position, respectively.

[0088] b) Set the y-axis of the coordinate system of a point on the seabed surface to be in the same direction as the side-scan sonar track line. At this time, the incident deflection angle τ of the sonar signal is 0.

[0089] c) According to the working mechanism of the side scan sonar, the corresponding inclination angle of the pixel at the xth position in the i-th frame of the side scan sonar image can be obtained: for:

[0090]

[0091] Where: arctan(·) is the inverse tangent function; N tra is the lateral pixel distance of the side scan sonar image; W i is the position of the seabed point in the i-th frame of the sidescan sonar image.

[0092] Step 3. Obtain the elevation of the 3D reconstructed surface.

[0093] The method of obtaining the three-dimensional reconstructed surface elevation includes inverting the seabed surface elevation value and designing the inverted elevation constraint coefficient k. i . The three-dimensional reconstruction of the side-scan sonar image SFS adopts a linearization method, that is, firstly, the radiance equation is approximated by a finite difference operation; then the radiance equation is discretized; and then the three-dimensional reconstructed surface elevation is calculated using the Newton-Raphson method. The linearization of the three-dimensional reconstruction of the side-scan sonar image SFS is to linearize the complex seabed surface reflection model function of the present invention, thereby converting the original SFS nonlinear problem into a linear problem for solution.

[0094] The inversion of the seafloor surface elevation value specifically adopts the following steps:

[0095] (I) The seafloor surface gradients p and q are discretized approximately as follows using the backward finite difference method:

[0096]

[0097] Where: Z(x,y) is the elevation of the seafloor surface at position (x,y); p and q are the horizontal and vertical gradients of the seafloor surface, respectively.

[0098] (II) The radiation illumination equation of the seabed surface in the side scan sonar image can be written as:

[0099] f[Z(x,y)]=0≈L(x,y)-R(p(x,y),q(x,y))

[0100] (III) After Taylor expansion of a pixel point (x, y) on the side scan sonar image, we can get:

[0101]

[0102] (IV) Let Z(x,y)=Z n (x,y), then the surface elevation value of the nth iteration is:

[0103]

[0104] Where: Z n (x,y),Z n-1 (x, y) are the surface elevation iteration results of the nth and n-1th times respectively.

[0105] (V) Set the seabed surface elevation iteration termination condition μ, that is, when the following expression is satisfied, the inversion of the seabed surface elevation value is completed.

[0106] Z n(x,y)-Z n-1 (x,y)≤μ

[0107] Where: μ is the iteration termination condition of the seabed surface elevation.

[0108] At present, the three-dimensional reconstruction method of SFS side-scan sonar images can only obtain the relative topography of the seabed, but cannot obtain the absolute topography of the seabed. For this reason, relevant researchers use external multi-beam bathymetric data to constrain the three-dimensional reconstructed terrain. External bathymetric data requires the use of multi-beam side-scan sonar to measure the same area of ​​the seabed, which reduces the efficiency of three-dimensional reconstruction to a certain extent, and excessive reliance on external bathymetric data will reduce the effect of the reflection model. For this reason, the present invention no longer relies on external water depth data to constrain the inverted terrain, but uses seabed point data extracted from side-scan sonar images to constrain the inverted terrain. The seabed point corresponds to the depth value of the current side-scan sonar from the seabed. After obtaining the positions of all seabed points, the depth data on the AUV track line can be obtained.

[0109] The design inversion elevation constraint coefficient k i , specifically the following steps:

[0110] (A) Set the seabed point position of the i-th frame of the side scan sonar image to W i , then the current depth of the AUV from the seabed is H i for:

[0111]

[0112] Where: U is the single-side scan width of the side scan sonar; N tra is the horizontal pixel distance of the side scan image.

[0113] (B) Design the following inversion elevation constraint coefficient k for the i-th frame of the side scan sonar image: i :

[0114]

[0115] Where: It is the inverted elevation value of the track line position of the i-th frame of the sidescan sonar image.

[0116] The three-dimensional reconstruction surface elevation is obtained by specifically adopting the following steps:

[0117] (a) Inverted seabed surface elevation values ​​from sidescan sonar images.

[0118] (b) Find all the depth data on the track line, then the depth set H = {H1, H2…, H i …,H o}.

[0119] (c) Calculate the inversion elevation constraint coefficient k of the i-th frame of the side scan sonar imagei .

[0120] (d) The expression of the inverted elevation value Φ(Z(x,i)) corresponding to the i-th frame of the constrained side scan sonar image is as follows:

[0121]

[0122] Where: Z(x,i) is the inverted elevation value corresponding to the i-th frame of the sidescan sonar image.

[0123] (e) After all frames of the i-th side-scan sonar image have been constrained in their corresponding elevation values, the elevation values ​​of other frames of the side-scan sonar image are constrained in turn, thereby completing the elevation scale constraint of the seabed surface of the side-scan sonar image.

[0124] Step 4. Output the 3D reconstruction result of the side-scan sonar image.

[0125] In order to verify the effectiveness of the three-dimensional reconstruction method of side-scan sonar images of the present invention, side-scan sonar images of seabed terrain and side-scan sonar images of underwater wind turbine pile foundations were selected for three-dimensional reconstruction verification, and the method of the present invention was compared with the Lambert model method and the Ward hybrid model method.

[0126] Attached comparative literature:

[0127] [1] Wang Ai-xue, Zhao Jian-hu, Shang Xiao-dong, et al. Three-dimensional micro-topography inversion from single-beam side-scan sonar images with water depth constraints[J]. Journal of Harbin Engineering University, 2017, 38(05): 739-745.

[0128] [2] Wang Guohui, Song Yugui, Ma Fengjun. Research progress on shape recovery from light and shade on non-Lambert surfaces[J]. Tool Technology, 2012, 46(08): 3-7.

[0129] Figures 3 to 6 The comparison results of different 3D reconstruction methods for four sets of side-scan sonar images are given, such as Figure 3 (a) and Figure 4 As shown in (a), the terrain of the two underwater terrain side-scan sonar images is relatively scattered. Figure 5 (a) and Figure 6 (a) shows two side-scan sonar images of underwater wind turbine pile foundations, where the pile foundations and their surrounding targets are relatively concentrated.

[0130] Depend on Figure 3 (b) Figure 4 (b) Figure 5 (b) and Figure 6It can be seen from (b) that based on the inversion of a single diffuse reflection model, the terrain features are blurred, and there are many abnormal protrusions caused by noise, which cannot accurately show the relative undulation of the original terrain. This is mainly because the commonly used SFS method uses a single Lambert diffuse reflection model for inversion, which does not match the actual seabed surface, causing the corresponding features of the two-dimensional side-scan image to be blurred in the inverted terrain. At the same time, new noise will also be introduced, resulting in abnormal protrusions in the inverted terrain, making it impossible to accurately reflect the changes in the actual seabed surface.

[0131] Depend on Figure 3 (c) Figure 4 (c) Figure 5 (c) and Figure 6 (c) shows that the terrain features reconstructed based on the Ward mixed reflection model are fuzzy, and there are obvious abnormalities in the terrain undulation in some local areas. There are also many abnormal convex points on the reconstructed seabed surface. This is because the Ward mixed reflection model tends to reflect on surfaces with less roughness and cannot adapt to complex seabed surfaces. In addition, since the parameters of the Ward mixed reflection model have a great influence on the inversion results, setting the parameters based on experience cannot achieve ideal 3D reconstruction effects, which in turn causes abnormal 3D reconstructed terrain undulation, which is quite different from the side scan sonar image features.

[0132] Compared with the above two 3D reconstruction methods, Figure 3 (d) Figure 4 (d) Figure 5 (d) and Figure 6 The reconstruction result of the method of the present invention shown in (d) clearly shows the changes in the height fluctuations and terrain features of the seabed surface, and there are no abnormal protrusions. It can effectively and intuitively express the seabed terrain information in the side-scan sonar image. This is mainly due to the fact that the method of the present invention combines the characteristics of the seabed surface near the continental shelf, designs the SFS mixed reflection model of the side-scan sonar image, no longer uses the diffuse reflection Lambert model alone, but introduces the Blinn-Phong specular reflection model, and uses the exponential function form to design the adaptive specular reflection coefficient D g , using the fraction containing the roughness C of the side scan sonar image to express D g The roughness C is used as the index to improve the correlation between the side scan sonar image and the seabed surface reflection model. In addition, an adaptive specular reflection index ε is designed to convert D g As the formula coefficient of the specular reflection index ε, the inverse of the roughness multiplied by the constant λ is used to express ε, and the adaptive diffuse reflection coefficient D is designed. r, forming an SFS mixed reflection model with complex seabed surface adaptation, improving the adaptability of the SFS mixed reflection model, thereby improving the accuracy of three-dimensional reconstruction of side scan sonar images. Compared with other three-dimensional reconstruction methods, the present invention designs the inversion elevation constraint coefficient k i , the depth data set H = {H1, H2…, H i …,H o}, and use the depth data and track line position to invert the elevation value Z(N tra / 2,i) to design the inversion height constraint coefficient k i , with this coefficient k i The inversion elevation value Z(x,i) is constrained to obtain the absolute elevation value Φ(Z(x,i)), so that the three-dimensional inversion terrain features of the present invention are clear, the fluctuation height changes are consistent with the characteristics of the seabed, and there are no abnormal protrusions, thereby achieving a high degree of matching and accurate three-dimensional reconstruction of the seabed terrain surface.

[0133] In order to further evaluate the advantages and disadvantages of the 3D reconstruction method of the present invention and the comparison method, the present invention introduces indicators such as mean absolute error (MAE), correlation coefficient (CC), and signal-to-noise ratio (SNR). The mean absolute error (MAE) is mainly used to evaluate the error between the 3D projection image and the original image, and the smaller the MAE value, the higher the similarity between the two, indicating that the 3D reconstruction of the image is more reasonable.

[0134]

[0135] Where: m, n are the pixel distances corresponding to the length and width of the image respectively; G z (i,j),G o (i, j) are the pixel values ​​of the three-dimensional projection image and the original image at position (i, j) respectively.

[0136] The correlation coefficient (CC) is mainly used to evaluate the similarity between the 3D projection image and the original image, reflecting the ability of the 3D reconstruction algorithm to discover the terrain information of the side scan sonar image. The larger the CC value, the better. The specific expression is as follows:

[0137]

[0138] Where: are the average pixel values ​​of the 3D projection image and the original image respectively.

[0139] The signal-to-noise ratio (SNR) is mainly used to evaluate the degree to which the 3D reconstruction algorithm is affected by image noise. The larger the SNR value, the less the 3D reconstruction algorithm is affected by image noise, and the better its performance.

[0140]

[0141] Where: lg is the logarithm with base 10.

[0142] The comparison of the three-dimensional reconstruction results of four side-scan sonar images is shown in Table 1. The MAE index of the SFS mixed reflection model three-dimensional reconstruction method in the present invention is reduced by 25.302% and 39.053% respectively compared with the Lambert reflection model method and the Ward mixed reflection model method; the CC index is increased by 9.590% and 49.649% respectively; the SNR index is increased by 23.004% and 31.456% respectively, which is mainly due to the SFS mixed reflection model of the side-scan sonar image designed by the present invention. The model introduces the Blinn-Phong specular reflection model to describe the specular reflection part of the seabed, forming an SFS mixed reflection model with complex seabed surface adaptation, which enhances the matching of the reflection model with the seabed surface. The present invention adopts an exponential function form to design the adaptive specular reflection coefficient D g , using the fraction containing the roughness C of the side scan sonar image to express D g The base number is the roughness C, and the adaptive diffuse reflectance coefficient D is designed. r The adaptive mirror reflection index ε improves the correlation between the side scan sonar image and the seabed surface reflection model, thereby reducing the MAE index of the method of the present invention and improving the accuracy of the three-dimensional reconstruction of the side scan sonar image. i , with this coefficient k i The constrained inversion elevation value Z(x,i) is used to obtain the absolute elevation value Φ(Z(x,i)), which reduces the dependence of the three-dimensional reconstruction of the side-scan sonar image on the external side-scan data, enhances the role of the mixed reflection model in the three-dimensional reconstruction, makes the three-dimensional reconstruction terrain more reasonable, and avoids the introduction of new noise in the three-dimensional reconstruction process. Therefore, the SNR index of the method of the present invention is improved, thereby improving the efficiency and accuracy of the three-dimensional reconstruction of the side-scan sonar image. As can be seen from the above, the comparison of the three-dimensional reconstruction results of the four side-scan sonar images further verifies that the method of the present invention can effectively achieve a high degree of matching and accurate three-dimensional reconstruction of the seabed terrain surface.

[0143] Table 1

[0144]

Claims

1. A method for three-dimensional reconstruction of side-scan sonar images, characterized in that: The steps include: Step (1): reading the side scan sonar image to be three-dimensionally reconstructed; Step (2): Design the SFS mixed reflection model of the side scan sonar image; The side scan sonar image SFS mixed reflection model is designed; including designing an adaptive mirror reflection coefficient D g ; Design adaptive diffuse reflectance D r ; Design an adaptive specular reflection index ε; Design an adaptive mixed reflection model expression for the seafloor surface; Step (3): Obtain the elevation of the three-dimensional reconstructed surface; The method of obtaining the three-dimensional reconstructed surface elevation includes inverting the seabed surface elevation value and designing the inverted elevation constraint coefficient k. i ; Step (4): Output the three-dimensional reconstruction result of the side-scan sonar image.

2. The method for three-dimensional reconstruction of side-scan sonar images according to claim 1, characterized in that: In step (2), the design of the adaptive mirror reflection coefficient D g , specifically including the following steps: 1) Calculate the surface roughness C of the side scan sonar image; 2) Adaptive mirror reflection coefficient D g , using the following formula:

3. The method for three-dimensional reconstruction of side-scan sonar images according to claim 1, characterized in that: In step (2), the design of the adaptive diffuse reflectance coefficient D r , using the following formula:

4. The method for three-dimensional reconstruction of side-scan sonar images according to claim 1, characterized in that: In step (2), the design of the adaptive mirror reflection index ε is specifically based on the following formula: Where: λ is a constant.

5. The method for three-dimensional reconstruction of side-scan sonar images according to claim 1, characterized in that: In step (2), the expression of the seabed surface adaptive mixed reflection model is designed, and the specific formula is as follows: Where: R hs (p,q) is the adaptive mixed reflection model function of the seabed surface, p and q are the horizontal and vertical gradients of the seabed surface, respectively. τ is the inclination and deflection of the incident direction of the sonar signal, ρ is the intensity of the incident sound wave, and p s ,q s are the horizontal and vertical gradients of the energy source position, respectively.

6. The method for three-dimensional reconstruction of side-scan sonar images according to claim 5, characterized in that: In designing the expression of the seabed surface adaptive mixed reflection model, the following steps are used to determine the parameters τ: a) Set the y-axis of the coordinate system of a point on the seabed surface to be in the same direction as the side-scan sonar track line, and the incident deflection angle τ of the side-scan sonar signal is 0; b) According to the working mechanism of side-scan sonar, the corresponding inclination angle of the pixel at position x in the i-th frame of the side-scan sonar image can be obtained: for: Where: arctan(·) is the inverse tangent function; W i is the position of the seabed point in the i-th frame of the side scan sonar image; c) Calculate the inclination angle of all pixels in each frame of the side-scan sonar image, and then obtain the corresponding inclination angle of the pixels of the entire side-scan sonar image.

7. The method for three-dimensional reconstruction of side-scan sonar images according to claim 1, characterized in that: In step (3), the inversion of the seafloor surface elevation value specifically adopts the following steps: (A) The backward finite difference method is used to discretize the seafloor surface gradients p and q approximately as follows: Where: Z(x,y) is the elevation of the seafloor surface at position (x,y); p and q are the horizontal and vertical gradients of the seafloor surface, respectively; (B) The radiation illumination equation of the seabed surface in the side scan sonar image can be written as: f[Z(x,y)]=0≈L(x,y)-R(p(x,y),q(x,y)) Where R(·) represents the reflection model function; (C) Taylor expansion of a pixel point (x, y) on the side scan sonar image yields: (D) Let Z(x,y)=Z n (x,y), then the surface elevation value of the nth iteration is: Where: Z n (x,y),Z n-1 (x, y) are the surface elevation iteration results of the nth and n-1th times respectively; (E) Setting the seafloor surface elevation iteration termination condition μ, that is, when the following expression is satisfied, the inversion of the seafloor surface elevation value is completed; Z n (x,y)-Z n-1 (x,y)≤μ Where: μ is the iteration termination condition of the seabed surface elevation.

8. The method for three-dimensional reconstruction of side-scan sonar images according to claim 1, characterized in that: In step (3), the design inversion elevation constraint coefficient k i , specifically the following steps: (I) Set the seabed point position of the i-th frame of the side scan sonar image to W i , then the current depth of the AUV from the seabed is H i for: Where: U is the single-side scan width of the side scan sonar; N tra is the lateral pixel distance of the side scan image; (II) Design the following inversion elevation constraint coefficient k for the i-th frame of the side scan sonar image: i : Where: It is the inverted elevation value of the track line position of the i-th frame of the sidescan sonar image.

9. The method for three-dimensional reconstruction of side-scan sonar images according to claim 1, characterized in that: In step (3), the three-dimensional reconstructed surface elevation is obtained by specifically adopting the following steps: (a) Invert the seabed surface elevation of the side scan sonar image; (b) Find all the depth data on the track line, then the depth set H = {H1, H2…, H i …}; (c) Calculate the inversion elevation constraint coefficient k of the i-th frame of the side scan sonar image i ; (d) The expression of the inverted elevation value Φ(Z(x,i)) corresponding to the i-th frame of the constrained side scan sonar image is as follows: Where: Z(x,i) is the inverted elevation value corresponding to the i-th frame of the side scan sonar image; (e) After all frames of the i-th side-scan sonar image have been constrained in their corresponding elevation values, the elevation values ​​of other frames of the side-scan sonar image are constrained in turn, thereby completing the elevation scale constraint of the seabed surface of the side-scan sonar image.

Citation Information

Cited By

  • Wind power pile foundation scouring detection method based on side scan sonar

    CN121810643A