Marine topographic survey method and system for multi-source heterogeneous data fusion processing

Abstract

The invention relates to a marine topographic measurement method and system for multi-source heterogeneous data fusion processing, and the method comprises the steps: obtaining the marine topographic data of different to-be-measured regions, and obtaining the multi-source heterogeneous data; preprocessing the multi-source heterogeneous data to obtain preprocessed data; performing feature extraction on the preprocessed data, and determining the weight of each data source based on the extracted features; performing fusion processing on the data of the plurality of data sources to obtain a corresponding fusion result; evaluating the fusion result according to the comprehensive evaluation index, and taking the first fusion result passing the evaluation as target ocean topographic data; and constructing an iterative optimization function, carrying out iterative optimization on the second fusion result which does not pass the evaluation on the comprehensive evaluation index based on the iterative optimization function until the optimized second fusion result passes the evaluation, and taking the optimized second fusion result which passes the evaluation as a target fusion result. According to the invention, the accuracy and the data processing effect of ocean topographic survey can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of ocean mapping technology, and in particular to a multi-source heterogeneous data fusion processing ocean topography measurement method and system, electronic equipment and non-transitory computer readable storage medium. BACKGROUND

[0002] Currently, ocean topography measurement methods mainly rely on a single data source, such as sonar measurement, satellite remote sensing or underwater laser scanning. These technologies have their own advantages and disadvantages. Sonar measurement can provide high-precision underwater topographic data, but is easily disturbed in complex marine environments; satellite remote sensing technology has a wide coverage range, but has low accuracy in underwater measurement; underwater laser scanning can obtain high-resolution three-dimensional data, but its scope of application is limited by water transparency and depth. Therefore, a single technology often cannot fully and accurately reflect the complexity of the ocean topography.

[0003] However, the existing approach has the disadvantage of strong dependence on data sources, resulting in limitations and uncertainties in measurement results. In complex marine environments, a single data source often cannot fully capture the subtle changes and diversity of the terrain. In addition, the fusion processing between different data sources has not been fully appreciated, and there is a lack of effective methods to integrate and utilize data from different technologies, thereby affecting the overall accuracy and reliability of ocean topography measurement. SUMMARY

[0004] The present application provides a multi-source heterogeneous data fusion processing ocean topography measurement method, system, electronic equipment and non-transitory computer readable storage medium that can improve the accuracy of ocean topography measurement and data processing effect.

[0005] The technical solution of the present application to solve the above technical problems is as follows:

[0006] The present application provides a multi-source heterogeneous data fusion processing ocean topography measurement method, the method comprising:

[0007] Obtain ocean topography data of different to-be-measured regions through multiple data sources to obtain multi-source heterogeneous data;

[0008] Preprocess the multi-source heterogeneous data to obtain preprocessed data;

[0009] Extract features from the preprocessed data, and determine the weight of each data source based on the extracted features;

[0010] According to the weight of each data source, the data of multiple data sources is fused to obtain the corresponding fusion result;

[0011] The fusion result is evaluated according to the comprehensive evaluation index, and a first fusion result passing the evaluation is taken as the target marine terrain data;

[0012] An iterative optimization function is constructed, and the second fusion result failing the evaluation is iteratively optimized based on the iterative optimization function until the optimized second fusion result passes the evaluation, and the optimized second fusion result passing the evaluation is taken as the target fusion result.

[0013] Optionally, the marine terrain data of different to-be-tested regions is acquired through multiple data sources to obtain multi-source heterogeneous data, including:

[0014] The first data of a deep water region is collected by a multi-beam sonar system carried by an unmanned ship;

[0015] The second data of a shallow water region is collected by a hyperspectral camera carried by an unmanned aerial vehicle;

[0016] A local high-precision third data is acquired by deploying an underwater laser scanner in a key region;

[0017] The first data, the second data and the third data are acquired to obtain the multi-source heterogeneous data.

[0018] Optionally, the multi-source heterogeneous data is preprocessed to obtain preprocessed data, including:

[0019] Noise in the multi-source heterogeneous data is filtered out by an adaptive wavelet transform method to obtain first intermediate data;

[0020] The first intermediate data is subjected to geometric correction and radiometric correction to obtain second intermediate data;

[0021] The second intermediate data of different data sources is unified to the same coordinate system and resolution to obtain the preprocessed data.

[0022] Optionally, the preprocessed data is subjected to feature extraction, and the weight of each data source is determined based on the extracted features, including:

[0023] The water depth data and the water turbidity data collected by the data source are extracted from the preprocessed data;

[0024] Data quality factors, first adjustment coefficients, second adjustment coefficients, third adjustment coefficients and fourth adjustment coefficients corresponding to the data source are acquired;

[0025] The weight of the data source is determined according to the data quality factors, the first adjustment coefficients, the second adjustment coefficients, the third adjustment coefficients and the fourth adjustment coefficients corresponding to the data source, and the water depth data and the water turbidity data collected by the data source.

[0026] Optionally, the data of the plurality of data sources is fused according to the weight of each data source to obtain a corresponding fusion result, comprising:

[0027] Obtaining an observation value, a local terrain curvature and a slope of each data source;

[0028] Adjusting the observation value, the local terrain curvature and the slope of each data source according to a first smoothing factor and a second smoothing factor to obtain the fusion result.

[0029] Optionally, the fusion function is represented as:

[0030]

[0031] Wherein, F(x, y, z) is a result of the fusion function, U i is an observation value of the i th data source, H is a local terrain curvature, S is a slope, μ and ρ are respectively a first smoothing factor and a second smoothing factor.

[0032] Optionally, the fusion result is evaluated according to a comprehensive evaluation index, and a first fusion result passing the evaluation is taken as the target marine terrain data, comprising:

[0033] Obtaining a fusion result corresponding to each sampling point of each data source and a true value corresponding to the fusion result;

[0034] Obtaining a data redundancy of the fusion result;

[0035] Processing a terrain gradient in the fusion result according to a first correction coefficient, and processing the data redundancy according to a second correction coefficient, and combining the fusion result and the true value to determine a corresponding comprehensive evaluation index;

[0036] If the comprehensive evaluation index of the fusion result is greater than a preset threshold, the fusion result is determined as the first fusion result passing the evaluation.

[0037] Optionally, the comprehensive evaluation index is represented as:

[0038]

[0039] Wherein, E is a comprehensive evaluation index, P i is a fusion result corresponding to an i th sampling point of each data source, T i is a true value of the fusion result of the i th sampling point, is a terrain gradient, R is a data redundancy, ∈ is a first correction coefficient, θ is a second correction coefficient, and n is a total number of sampling points.

[0040] Optionally, the iterative optimization based on the comprehensive evaluation index on the second fusion result that fails to pass the evaluation comprises:

[0041] obtaining a preset learning rate, a convergence control parameter, and a Laplacian of an optimization variable of a current iteration round;

[0042] obtaining a first fusion function value of the current iteration round and a second fusion function value of a previous iteration round of the current iteration round;

[0043] performing iterative optimization on the first fusion function value and the second fusion function value according to the learning rate, the convergence control parameter, and the Laplacian to obtain an optimized second fusion result.

[0044] The application further provides a marine topographic measurement system for multi-source heterogeneous data fusion processing, which comprises:

[0045] a data acquisition module configured to acquire marine topographic data of different to-be-measured regions through multiple data sources to obtain multi-source heterogeneous data;

[0046] a data processing module configured to pre-process the multi-source heterogeneous data to obtain pre-processed data;

[0047] a weight determination module configured to extract features from the pre-processed data and determine weights of each data source based on the extracted features;

[0048] a data fusion module configured to fuse data of the multiple data sources according to the weights of each data source to obtain corresponding fusion results;

[0049] a first processing module configured to evaluate the fusion results according to a comprehensive evaluation index and take a first fusion result that passes the evaluation as target marine topographic data;

[0050] a second processing module configured to construct an iterative optimization function and perform iterative optimization on a second fusion result that fails to pass the evaluation based on the comprehensive evaluation index until an optimized second fusion result that passes the evaluation is obtained, and take the optimized second fusion result that passes the evaluation as target fusion result.

[0051] In addition, to achieve the above object, the application further provides an electronic device, which comprises a memory configured to store a computer software program and a processor configured to read and execute the computer software program, thereby realizing the marine topographic measurement method for multi-source heterogeneous data fusion processing as described above.

[0052] In addition, in order to achieve the above object, the application further provides a non-transitory computer readable storage medium, wherein the storage medium stores a computer software program, and the computer software program is executed by a processor to realize the multi-source heterogeneous data fusion processing ocean topographic measurement method.

[0053] The application has the following beneficial effects:

[0054] (1) The application adopts multi-beam sonar, hyperspectral remote sensing and underwater laser scanning three data acquisition means, and realizes all-around and fine coverage of deep water, shallow water and local area through the cooperative work of unmanned boats, unmanned aerial vehicles and laser scanners. Compared with the traditional single data source ocean measurement, the multi-source data acquisition makes the measurement in different sea environments more comprehensive, and ensures the integrity and accuracy of the topographic data of each region.

[0055] (2) The application realizes high-precision multi-source data integration by defining weight coefficients and fusion functions and effectively weighting and fusing data of different sources. According to the factors such as water depth and turbidity, the weight is dynamically adjusted, so that the data contribution value of different regions is rationalized, which helps to reduce the deviation between data and improve the accuracy of the fusion result.

[0056] (3) The application performs precision evaluation on the fusion result through the set comprehensive evaluation index, and dynamically corrects according to the factors such as data redundancy and topographic gradient. Compared with the static measurement method, the adaptive optimization mechanism significantly enhances the adaptability to data complexity and regional differences, realizes the dynamic balance of accurate measurement, and can adapt to the changing ocean environment.

[0057] In summary, the application realizes a better balance between wide-area coverage and local fine through the cooperative acquisition and adaptive fusion of multi-source data, can significantly improve the overall precision and measurement efficiency of ocean topographic measurement, and becomes the preferred scheme for fine topographic measurement in complex sea environment. BRIEF DESCRIPTION OF DRAWINGS

[0058] Figure 1 A scene diagram of the multi-source heterogeneous data fusion processing ocean topographic measurement method provided by the application;

[0059] Figure 2 A flowchart of the multi-source heterogeneous data fusion processing ocean topographic measurement method provided by the application;

[0060] Figure 3 A structural schematic diagram of the multi-source heterogeneous data fusion processing ocean topographic measurement system provided by the application;

[0061] Figure 4A possible hardware structure schematic diagram of an electronic device provided by the present application is shown in the following figure.

[0062] Figure 5 A possible hardware structure schematic diagram of a computer readable storage medium provided by the present application is shown in the following figure. DETAILED DESCRIPTION

[0063] The technical solutions in the embodiments of the present application will be clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all the other embodiments obtained by a person skilled in the art without creative work fall within the protection scope of the present application.

[0064] In the description of the present application, the terms "first", "second" are used only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0065] In the description of the present application, the term "for example" is used to indicate "as an example, illustration or explanation". Any embodiment described as "for example" in the present application is not necessarily interpreted as more preferred or more advantageous than other embodiments. The following description is given in order to enable any person skilled in the art to implement and use the present application. In the following description, details are listed for the purpose of explanation. It should be understood that a person skilled in the art can realize the present application without using these specific details. In other examples, well-known structures and processes will not be described in detail in order to avoid unnecessary details making the description of the present application obscure. Therefore, the present application is not intended to be limited to the shown embodiments, but is consistent with the broadest scope in accordance with the principles and characteristics disclosed.

[0066] Please refer to Figure 1 , Figure 1 A scene diagram of a marine topographic survey method of multi-source heterogeneous data fusion processing provided by the present application is shown in the following figure. As shown in the figure, the terminal and the server are connected through a network, such as a wired or wireless network connection. Among them, the terminal can include but is not limited to mobile phones, tablets and other portable terminals installed with various network platform applications, as well as computers, inquiry machines, advertising machines and other fixed terminals. Among them, the server provides various service services for users, including service push servers, user recommendation servers, etc. Figure 1

[0067] It should be noted that Figure 1 ​The scene diagram of the ocean topographic measurement method of the multi-source heterogeneous data fusion processing shown is only an example, and the terminal, server and application scenario described in the embodiments of the application are used to more clearly illustrate the technical solutions of the embodiments of the application, and do not limit the technical solutions provided by the embodiments of the application. It is known to those skilled in the art that with the evolution of the system and the appearance of new business scenarios, the technical solutions provided by the embodiments of the application are also applicable to similar technical problems.

[0068] The terminal can be used for:

[0069] Obtaining ocean topographic data of different to-be-measured regions through multiple data sources to obtain multi-source heterogeneous data;

[0070] Preprocessing the multi-source heterogeneous data to obtain preprocessed data;

[0071] Extracting features from the preprocessed data, and determining the weight of each data source based on the extracted features;

[0072] According to the weight of each data source, the data of multiple data sources is fused to obtain a corresponding fusion result;

[0073] According to the comprehensive evaluation index, the fusion result is evaluated, and the first fusion result that passes the evaluation is taken as the target ocean topographic data;

[0074] An iterative optimization function is constructed, and the second fusion result that fails the evaluation is iteratively optimized based on the iterative optimization function until the optimized second fusion result passes the evaluation, and the optimized second fusion result that passes the evaluation is taken as the target fusion result.

[0075] Please refer to Figure 2 , a flowchart of the ocean topographic measurement method of the multi-source heterogeneous data fusion processing of the application is provided, including the following steps:

[0076] Step 201, obtaining ocean topographic data of different to-be-measured regions through multiple data sources to obtain multi-source heterogeneous data.

[0077] In some embodiments, step 201 can include:

[0078] Using a multi-beam sonar system carried by an unmanned boat to collect first data of a deep water region;

[0079] Collecting second data of a shallow water region by an unmanned aerial vehicle carrying a hyperspectral camera;

[0080] Deploying an underwater laser scanner in a key area to obtain third data of local high precision;

[0081] Obtaining the first data, the second data and the third data, and obtaining the multi-source heterogeneous data.

[0082] Wherein, the multi-beam sonar system is a detection device suitable for deep water environment, which measures underwater terrain by transmitting and receiving multiple acoustic beams. It has strong penetration and can penetrate the water layer in deep water area to provide accurate terrain depth data. The unmanned boat carries the multi-beam sonar and can operate autonomously in deep water area, reducing the need for manual operation, and is particularly suitable for continuous and large-scale terrain mapping tasks in deep water area. The first data mainly includes terrain depth and profile data in deep water area, which is suitable for large-scale deep water terrain modeling.

[0083] Wherein, the hyperspectral camera has rich spectral information acquisition capability and can distinguish the reflection characteristics of different substances in the shallow layer of water. In shallow water area, optical data has good penetration ability, and hyperspectral camera can collect detailed information of underwater shallow terrain and material distribution. The flexibility of unmanned aerial vehicle is high, which can fly close to the shallow water area, so as to obtain the terrain and water quality data of the shallow layer of water with high resolution. This method is suitable for shallow water area with complex terrain, and reduces optical distortion and interference by low-altitude flight. The second data includes surface terrain, material distribution information and spectral feature data in shallow water area, which is suitable for fine shallow water terrain mapping and environmental analysis.

[0084] Wherein, the underwater laser scanner is a high-precision detection device that can collect micro-terrain data in local underwater area. It is usually used for high-precision data collection in specific areas, such as measuring seabed structure, rock fault or sediment characteristics. The third data can include high-precision terrain data in local area, which is suitable for detailed analysis and three-dimensional reconstruction, and can be used for further processing and model verification.

[0085] It can be understood that the first data, the second data and the third data obtained by the above three devices form multi-source heterogeneous data. These data contain marine terrain information of different depths (deep water, shallow water) and different precisions (large range, medium precision, local high precision). Through weight calculation and data fusion model processing, multi-source heterogeneous data can overcome the limitations of single data source, improve the comprehensiveness and accuracy of terrain measurement. This process makes the terrain information from deep water to shallow water and then to local key area more complete, which helps to establish an accurate and hierarchical marine terrain model.

[0086] Through the above three collection methods and multi-source data fusion processing, the present application can obtain complete and accurate marine terrain data, which provides solid data support for fine terrain measurement and data modeling in complex sea areas.

[0087] Step 202, preprocessing the multi-source heterogeneous data to obtain preprocessed data.

[0088] In some embodiments, step 202 can include:

[0089] Filtering out the noise in the multi-source heterogeneous data by an adaptive wavelet transform method to obtain first intermediate data;

[0090] Performing geometric correction and radiometric correction on the first intermediate data to obtain second intermediate data;

[0091] Unifying the second intermediate data of different data sources to the same coordinate system and resolution to obtain the preprocessed data.

[0092] Wherein, the wavelet transform is a commonly used signal processing method, which is suitable for processing noise removal problems. The adaptive wavelet transform dynamically adjusts the filtering parameters according to the data characteristics to effectively remove random noise and interference signals in the data. Multi-source heterogeneous data (such as sonar data, hyperspectral data and laser scanning data) will be affected by environmental interference and equipment errors during acquisition. By wavelet transform denoising, the clarity of the data can be improved, and the data characteristics are more prominent. The first intermediate data obtained after denoising is the measurement data with greatly reduced noise, which lays a foundation for subsequent geometric and radiometric correction.

[0093] Wherein, the geometric correction can be used to eliminate the spatial deformation of the measurement data caused by the change of the viewing angle or the platform attitude, so that the geometric shape of the data is consistent with the actual terrain. This process restores the true position of the data through a mathematical model to ensure the accuracy of the data in space. The radiometric correction adjusts the spectral distortion of optical data (such as hyperspectral data) under different lighting conditions, so that it reflects the actual spectral intensity and distribution. This step can make the optical data maintain consistent spectral characteristics under different environments. The second intermediate data obtained after geometric and radiometric correction is more accurate in space and radiation accuracy, providing a basis for the unified processing of different data sources.

[0094] In specific implementation, different data sources can have different coordinate systems (such as WGS84, UTM, etc.). Converting all data to a unified coordinate system allows data to be overlaid on the same geographic space. The spatial resolution of different data sources may be different (for example, sonar data may be coarse, and hyperspectral and laser scanning data may be finer). Unifying the data sources to the same resolution can ensure the consistency of the data in spatial distribution, thereby avoiding the problem of inconsistent scales when fusion processing. After the coordinate system and resolution unification processing, the second intermediate data of all data sources form a set of standardized and preprocessed data, laying a foundation for the fusion of multi-source data.

[0095] In conclusion, the present application successfully eliminates the noise and spatial inconsistency in multi-source data through preprocessing steps such as adaptive wavelet denoising, geometric and radiometric correction, coordinate system and resolution unification, ensures the consistency and accuracy of the data in space. The preprocessed data can more accurately reflect the topographic features of different regions, and help the subsequent data fusion process to proceed smoothly, thereby improving the accuracy and reliability of the final terrain model.

[0096] Step 203, feature extraction is performed on the preprocessed data, and the weight of each data source is determined based on the extracted features.

[0097] In some embodiments, step 203 can include:

[0098] extracting water depth data and water turbidity data collected by the data source from the preprocessed data;

[0099] obtaining data quality factors, first adjustment coefficients, second adjustment coefficients, third adjustment coefficients and fourth adjustment coefficients corresponding to the data source;

[0100] determining the weight of the data source according to the data quality factors, the first adjustment coefficients, the second adjustment coefficients, the third adjustment coefficients and the fourth adjustment coefficients corresponding to the data source, and the water depth data and the water turbidity data collected by the data source.

[0101] In some embodiments, the weight of the data source can be expressed as:

[0102] W = a * exp(-β * D) * (1-γ * T) * (1+λ * Q);

[0103] where W is the weight of the data source, D represents water depth, T represents water turbidity, Q represents data quality factor, and a, β, γ, λ are the first adjustment coefficient, the second adjustment coefficient, the third adjustment coefficient and the fourth adjustment coefficient, respectively.

[0104] In a specific implementation, W represents the proportion or importance of a specific data source in the multi-source fusion process. A larger W value indicates that the data source has a higher weight in the fusion model, indicating that the data source has a greater impact on the final measurement result. The size of the weight coefficient is adjusted according to multiple factors such as water depth, turbidity and data quality, to ensure the rationality and accuracy of the data source in different environments.

[0105] In exp(-β * D), the greater the water depth, the higher the possibility of data source interference by environmental factors, especially optical and sonar data, which are severely attenuated in deep water. Therefore, this term is adjusted using a negative exponential function, and as the water depth D increases, the value of exp(-β * D) gradually decreases, gradually reducing the data weight in deep water areas.

[0106] The sensitivity of the weight decay to the water depth is adjusted by β. A larger β value accelerates the weight decay, which is suitable for unstable measurements in deep water. A smaller β value slows down the weight decay, which is suitable for shallow water.

[0107] In (1-γ·T), the water turbidity affects the clarity of the data source, especially for optical imaging. The higher the turbidity, the lower the underwater visibility, and the lower the data reliability. Therefore, the weight is linearly decayed by (1-γ·T) to ensure that the data weight in turbid water is low. As T increases, this term decreases, reducing the data weight.

[0108] The adjustment coefficient γ is used to adjust the degree of influence of water turbidity on data weight. A larger γ value makes the weight more sensitive to turbidity, which is suitable for shallow water measurement environments with high clarity requirements.

[0109] In 1+λ·Q, the data quality factor Q represents the objective quality or reliability of the data source, which may be determined by equipment accuracy, environmental adaptability, etc. The higher the quality factor, the more reliable the data, and the weight is also increased accordingly. Linear amplification is achieved by 1+λ·Q to increase the weight of high-quality data.

[0110] λ is used to control the influence of data quality factor on weight. A larger λ value enhances the role of high-quality data, which is suitable for measurement scenarios that require strict data quality assurance.

[0111] α is a global adjustment coefficient, which is used to standardize and unify the overall level of the weight. It is usually set according to actual measurement requirements to adjust the range of the weight. Adjusting the value of α can change the overall level of the weight, but it does not affect the relative role of D, T, and Q factors.

[0112] This formula combines water depth, turbidity, and data quality, taking into account the environmental adaptability and actual data quality of each data source. Through exponential decay and linear adjustment, each factor plays a reasonable role in different situations.

[0113] When the measurement environment changes (such as switching from shallow water to deep water), the weight coefficient will automatically adjust, reducing the weight of optical data in deep water and increasing the influence of high-quality data, thereby improving the accuracy of the fusion result.

[0114] By adjusting the comprehensive weight coefficient, the weight distribution of different data sources is reasonable, optimizing the overall fusion effect of multi-source data and adapting to the complex and variable marine environment measurement requirements.

[0115] If a data source measures deep water area (large D value), low turbidity (small T value), and high equipment quality factor Q, then the weight W of the data source will be reduced by the exp(-β·D) term, but amplified by the (1+λ·Q) term, forming a balance of the weight.

[0116] Conversely, in the case of shallow water area, high turbidity, and general data quality, the weight W will be reduced by the superposition of multiple factors, avoiding the negative impact of the data source on the fusion result.

[0117] In summary, the present application can realize effective screening and integration of multi-source data, so that the final fusion result has high reliability and adaptability in different sea environment.

[0118] Step 204, according to the weight of each data source, the data of multiple data sources are fused to obtain the corresponding fusion result.

[0119] In some embodiments, step 204 can include:

[0120] Obtaining the observation value, local terrain curvature and slope of each data source;

[0121] According to the first smoothing factor and the second smoothing factor, the observation value, the local terrain curvature and the slope of each data source are adjusted to obtain the fusion result.

[0122] In some embodiments, the fusion function is represented as:

[0123]

[0124] Where F(x, y, z) is the result of the fusion function, U i is the observation value of the i th data source, H is the local terrain curvature, S is the slope, and μ, ρ are the first smoothing factor and the second smoothing factor, respectively.

[0125] In a specific implementation, F(x, y, z) is the terrain result after fusion, representing the terrain value at a given point (x, y, z). By weighted average of multi-source data and introduction of terrain smoothing control and slope information, the result can represent the true terrain characteristics.

[0126] W i is the weight coefficient of the i th data source, representing the importance of each data source.

[0127] U i is the observation value of the i th data source, which is usually the marine terrain measurement value of the data source at a certain position in this scenario.

[0128] μ and ρ are smoothing factors, used to control the degree of smoothing and the slope influence of the fusion result, respectively adjusting the weight of terrain curvature and slope in the result.

[0129] is the second-order gradient of local terrain (i.e. Laplacian), used to measure the curvature of terrain. This term reflects the degree of curvature or fluctuation of the current terrain surface.

[0130] S is the slope information, representing the degree of inclination of the terrain, used to affect the strength of smoothing.

[0131] This part realizes the weighted average of different data sources. By observing the value U i weighted average is performed according to the weight (W i , so that the contribution of different data sources to the final fusion value is proportional to its weight. The significance of weighted average is that high-quality data (high weight) has a greater contribution to the final result, while data with more noise (low weight) has less impact. This weighting mechanism can improve the accuracy of data fusion and reduce errors caused by low-quality data.

[0132] Smooth control term This term is used to smooth the weighted average result, so that the fused terrain model is more reasonable in continuity and smoothness.

[0133] Local terrain curvature reflects the fluctuation of the terrain surface (e.g. ridges or valleys). The greater the curvature, the more dramatic the terrain changes. By multiplying the curvature term, areas with greater terrain fluctuations can be given stronger smoothing.

[0134] Slope attenuation term exp(-ρ·S), the greater the slope S, the more dramatic the terrain changes. At this time, the strength of smoothing needs to be reduced to preserve details. Therefore, by attenuating the smoothness through exp(-ρ·S), the smoothing factor decreases with the increase of slope. This ensures that in areas with large slopes, the terrain will not be over-smoothed, thus preserving the subtle terrain features.

[0135] The smoothing factor μ is used to control the overall influence of the smoothing term. A larger μ value will increase the strength of smoothing, making the terrain result more flat; while a smaller μ value will reduce the smoothing strength, more accurately reflecting the changes in the original terrain data.

[0136] The overall purpose of this formula is to generate a comprehensive terrain model based on the observation values of different data sources, considering the weighted contribution of multi-source data, and making the model have good continuity and naturalness through smoothing. The key mechanisms include:

[0137] Weighted fusion: By weighted average terms, ensure that different data sources have different effects on the results according to their quality and applicability, so as to improve the overall reliability of the fused data.

[0138] Adaptive smoothing: Introduce curvature and slope control smoothing processing, so that the terrain model remains clear in areas with steep slope changes, while providing natural transitions in flat areas, avoiding discontinuities or abrupt changes in the model.

[0139] In marine terrain measurement, if multiple data sources (such as sonar data, hyperspectral data and laser data) provide different observation values U i at a certain location, the formula will be weighted average according to the weight W i to ensure accurate fusion.

[0140] In areas with high or steep slope of the terrain, the smoothing term plays a control role, so that the data remains smooth without losing details.

[0141] The fusion function of the present application achieves a balance between data weighting and smoothing control, which can reasonably integrate multi-source data and adapt to terrain features, so as to generate an accurate model that conforms to the actual terrain features. By controlling the influence of each term, the formula provides an efficient and accurate data fusion method for complex marine terrain.

[0142] Step 205, according to the comprehensive evaluation index, evaluate the fusion result, and take the first fusion result that passes the evaluation as the target marine terrain data.

[0143] In some embodiments, step 205 can include:

[0144] Obtain the fusion result corresponding to each sampling point of each data source, and the true value corresponding to the fusion result;

[0145] Obtain the data redundancy of the fusion result;

[0146] According to the first correction coefficient, process the terrain gradient in the fusion result, and according to the second correction coefficient, process the data redundancy, and combine the fusion result and the true value to determine the corresponding comprehensive evaluation index;

[0147] If the comprehensive evaluation index of the fusion result is greater than a preset threshold, the fusion result is determined as the first fusion result that passes the evaluation.

[0148] In some embodiments, the comprehensive evaluation index is expressed as:

[0149]

[0150] where E is the comprehensive evaluation index, P i is the fusion result corresponding to the i-th sampling point of each data source, T i is the true value of the fusion result of the i-th sampling point, is the terrain gradient, R is the data redundancy, ∈ is the first correction coefficient, θ is the second correction coefficient, and n is the total number of sampling points.

[0151] In specific implementations, E is used for the comprehensive evaluation index, which is used to measure the overall accuracy and reliability of the fusion result. A smaller E value indicates that the error between the fusion result and the true value is small, and the reliability is high.

[0152] P i is the fusion result of each data source at the i-th sampling point, i.e., the measurement value calculated by the fusion model.

[0153] T i is the true value of the i-th sampling point, i.e., the reference value obtained in actual measurement.

[0154] is the terrain gradient, which is used to reflect the rate of change of the terrain surface. Generally, a larger value means that the terrain changes dramatically, and a smaller value indicates that the terrain is relatively flat.

[0155] R is the data redundancy, which refers to the amount of redundant data in the fusion result. The higher the redundancy, the more repeated sampling of data in the same area.

[0156] ∈ is the first correction coefficient, which is used to adjust the influence of the terrain gradient on the evaluation index.

[0157] θ is the second correction coefficient, which is used to adjust the influence of the data redundancy on the evaluation index.

[0158] This term is the Root Mean Square Error (RMSE), which is used to evaluate the deviation between the fusion result P i and the true value T i . By taking the average of the sum of squares of errors for all sampling points and then taking the square root, RMSE can better reflect the average level of error.

[0159] The RMSE term represents the basic fusion accuracy evaluation, which measures the deviation of the fusion model output data. The smaller the value, the closer the fusion result is to the true value. RMSE has strong sensitivity, and has higher weight for points with larger errors, which helps to identify areas with error anomalies.

[0160] The terrain gradient correction term According to the terrain gradient The RMSE result is modified. The larger the terrain gradient, the more dramatic the change in the terrain surface, which may lead to an increase in the difficulty of data fusion. By The value of the evaluation index E is increased, which reflects that in areas with dramatic changes, the error has a greater impact on the result.

[0161] The correction coefficient ∈ is used to control the influence of the terrain gradient on the evaluation index. A larger ∈ value will increase the correction effect of the terrain gradient, and is more suitable for scenarios that require strict control of complex terrain.

[0162] In areas with dramatic changes in terrain, the E value is increased to emphasize the accuracy requirements of the fusion result in steep or complex terrain, ensuring that the measured data adapts to complex terrain.

[0163] The data redundancy R in the redundancy decay term exp(-θ·R) represents the repetitiveness of the data. Higher redundancy usually indicates multiple acquisitions of the same terrain data, which can improve the stability of the fusion result. By using the negative exponential function exp(-θ·R) to decay the redundancy, this term tends to 0 when the redundancy is increased, which helps to reduce the E value, thereby reflecting the stability brought by high redundancy data. The correction coefficient θ controls the influence of redundancy on the evaluation index. A larger θ value will make the redundancy decay faster, making E decrease more significantly in the case of redundant data. This term encourages data redundancy and increases data stability, especially suitable for scenarios with a large amount of multi-source repeated data, making the fusion result more reliable in the case of increased redundant data.

[0164] The comprehensive evaluation index E integrates multiple factors to comprehensively evaluate the applicability of the fusion result from the aspects of accuracy, terrain complexity, and data stability. Specifically: the root mean square error term is used to evaluate the basic accuracy of data fusion. The terrain gradient correction term makes the accuracy requirements of complex terrain more stringent, reducing the loss of accuracy due to dramatic changes in terrain. The data redundancy correction term emphasizes the stability and reliability of the data, making high-redundancy data sources exhibit lower evaluation index values in the fusion result, i.e., higher accuracy.

[0165] If the terrain in a certain area is complex and has high redundancy, the terrain gradient is large, and the redundancy R is also high. In this case, the terrain gradient correction term will increase the E value, but due to the effect of the redundancy decay term exp(-θ·R), the overall E value may not increase significantly, thereby making the evaluation of this area more reflective of the actual accuracy.

[0166] In areas with gentle terrain and low redundancy, the gradient term and the redundancy term have less impact on E, and the evaluation of the fusion result is mainly dominated by the root mean square error term, making the evaluation index reflect the basic accuracy level.

[0167] The evaluation index E of the application combines error accuracy, terrain changes and data redundancy, and constructs a multi-factor adaptive evaluation method, so that the fusion result can better reflect the actual accuracy and applicability under different terrain conditions, and help to identify the reliability and data fusion effect difference in different regions.

[0168] Step 206, constructing an iterative optimization function, and iteratively optimizing the second fusion result which fails to pass the comprehensive evaluation index based on the iterative optimization function until the optimized second fusion result passes the evaluation, and taking the optimized second fusion result which passes the evaluation as the target fusion result.

[0169] In some embodiments, step 206 can include:

[0170] acquiring a preset learning rate, a convergence control parameter, and a Laplacian of an optimization variable of a current iteration round;

[0171] acquiring a first fusion function value of the current iteration round and a second fusion function value of a previous iteration round of the current iteration round;

[0172] iteratively optimizing the first fusion function value and the second fusion function value according to the learning rate, the convergence control parameter and the Laplacian to obtain an optimized second fusion result.

[0173] In some embodiments, the iterative optimization function can be represented as:

[0174]

[0175] wherein O(k+1) is an optimized variable of the iterative optimization function after the k+1th iteration, η is a learning rate, F(k) represents a first fusion function value of the current iteration, F(k-1) represents a second fusion function value of the last iteration, and -φ is a convergence control parameter, which represents a Laplacian of O(k).

[0176] In a specific implementation, O(k) is an optimized variable of the kth iteration, representing the fusion optimization value of the current iteration. O(k+1) is an optimized variable after the k+1th iteration, used to calculate the optimization result of the next time. η is a learning rate or a step factor, which controls the size of the update amount at each iteration. F(k) and F(k-1) represent the fusion function values of the current and last iterations respectively, which are used to measure the changes after the current data fusion. φ is a convergence control parameter, which is used to control the influence of the smoothing term on the iteration update step. is the Laplacian of O(k), that is, the second order gradient (or curvature) of the current terrain data optimization value, which is used to measure the smoothness of the current data.

[0177] In some embodiments, F(k)-F(k-1) reflects the change amplitude of the current iteration fusion value. Specifically, when the difference between F(k) and F(k-1) is large, it means that the current data fusion state is still far from the ideal state, and therefore a larger adjustment of O value is needed. When the difference between F(k) and F(k-1) is small, it means that the data fusion is gradually approaching stability, and the update amount should be gradually reduced to converge to an optimal solution.

[0178] This is used to control the update amplitude: The larger the absolute value of F(k)-F(k-1) is, the greater the curvature of the current optimization result is, indicating that there is a large discontinuity or drastic change in the data fusion result. In this case, the value of the exponential decay term will approach zero, thereby reducing the update amplitude to avoid unstable fluctuations. Correspondingly, when the value of F(k)-F(k-1) is small, the current optimization result is smooth, and the decay effect is small, so a larger update step can be maintained to improve the convergence speed. The smaller the value of F(k)-F(k-1) is, the smoother the current optimization result is, and the smaller the decay effect is, so a larger update step can be maintained to improve the convergence speed.

[0179] The learning rate η controls the basic size of each iteration step. If it is too large, the optimization may oscillate or diverge; if it is too small, the convergence speed may be too slow, increasing the computational cost.

[0180] It can be understood that the core purpose of this formula is to gradually adjust the optimization target based on the optimization value O(k) of the last iteration, combined with the change of the output result F of the data fusion model and the smoothness of the current optimization value, to make the fusion data result more accurate and smooth. The implementation process includes the following stages:

[0181] In the initial stage, when the iteration just starts, the difference term F(k)-F(k-1) is large, and O(k) has not been completely smoothed, so the formula will update with a larger step to quickly approach the target value. As the iteration deepens, when the curvature of O(k) gradually decreases and the smoothness increases, the exponential decay term gradually increases, making the adjustment step gradually decrease to achieve fine optimization. When F(k) and F(k-1) are almost equal and the value of F(k)-F(k-1) is small, the update amount tends to zero, and the optimization process gradually converges to obtain a smooth and accurate fusion result.

[0182] In summary, by introducing the control of the difference term and the Laplacian operator, the accuracy and convergence speed of the optimization are balanced, making the fusion result of multi-source heterogeneous data more stable and accurate, especially suitable for ocean measurement scenarios with complex terrain changes.

[0183] Please refer to Figure 3 , Figure 3 ​A structural schematic diagram of a marine topographic survey system provided by the present application for multi-source heterogeneous data fusion processing.

[0184] As shown in Figure 3 , the marine topographic survey system provided by the present application for multi-source heterogeneous data fusion processing comprises:

[0185] The data acquisition module 301 is configured to acquire marine topographic data of different to-be-measured regions through multiple data sources, and obtain multi-source heterogeneous data.

[0186] The data processing module 302 is configured to pre-process the multi-source heterogeneous data, and obtain pre-processed data.

[0187] The weight determination module 303 is configured to extract features from the pre-processed data, and determine the weight of each data source based on the extracted features.

[0188] The data fusion module 304 is configured to fuse the data of multiple data sources according to the weight of each data source, and obtain a corresponding fusion result.

[0189] The first processing module 305 is configured to evaluate the fusion result according to a comprehensive evaluation index, and take a first fusion result that passes the evaluation as target marine topographic data.

[0190] The second processing module 306 is configured to construct an iterative optimization function, and iteratively optimize a second fusion result that fails the evaluation based on the iterative optimization function until the optimized second fusion result passes the evaluation, and take the optimized second fusion result that passes the evaluation as target fusion result.

[0191] Please refer to Figure 4 , Figure 4 An embodiment schematic diagram of an electronic device provided by the present application is shown in the figure. Figure 4 As shown in the figure, the electronic device 400 provided by the present application comprises a memory 410, a processor 420, and a computer program 411 stored in the memory 410 and executable on the processor 420.

[0192] The marine topographic data of different to-be-measured regions is acquired through multiple data sources, and multi-source heterogeneous data is obtained.

[0193] The multi-source heterogeneous data is pre-processed, and pre-processed data is obtained.

[0194] Features are extracted from the pre-processed data, and the weight of each data source is determined based on the extracted features.

[0195] According to the weight of each data source, the data of the plurality of data sources is fused to obtain a corresponding fusion result;

[0196] According to the comprehensive evaluation index, the fusion result is evaluated, and a first fusion result that passes the evaluation is taken as target marine topographic data;

[0197] An iterative optimization function is constructed, and the second fusion result that fails the evaluation is iteratively optimized based on the iterative optimization function until the optimized second fusion result passes the evaluation, and the optimized second fusion result that passes the evaluation is taken as the target fusion result.

[0198] Please refer to Figure 5 , Figure 5 An embodiment of a computer readable storage medium provided by the embodiment of the present application is shown in the figure. Figure 5 As shown in the figure, the embodiment provides a computer readable storage medium 500, which stores a computer program 411, and the computer program 411 is executed by a processor to implement the following steps:

[0199] Marine topographic data of different to-be-measured areas is obtained through a plurality of data sources to obtain multi-source heterogeneous data;

[0200] The multi-source heterogeneous data is preprocessed to obtain preprocessed data;

[0201] The preprocessed data is feature extracted, and the weight of each data source is determined based on the extracted features;

[0202] According to the weight of each data source, the data of the plurality of data sources is fused to obtain a corresponding fusion result;

[0203] According to the comprehensive evaluation index, the fusion result is evaluated, and a first fusion result that passes the evaluation is taken as target marine topographic data;

[0204] An iterative optimization function is constructed, and the second fusion result that fails the evaluation is iteratively optimized based on the iterative optimization function until the optimized second fusion result passes the evaluation, and the optimized second fusion result that passes the evaluation is taken as the target fusion result.

[0205] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and the parts not described in detail in a certain embodiment can be referred to the related description of other embodiments.

[0206] Those skilled in the art will appreciate that embodiments of the present application can be devised for a variety of applications. It is intended that the present application be limited only by the scope of the appended claims, and it is intended that various modifications and alterations made by those skilled in the art be considered as within the scope of the present application. The embodiments of the present application will be described with reference to the attached drawings, wherein:

[0207] The present application is described in reference to the drawings using a flowchart and / or a block diagram of the method, apparatus (system) and computer program product according to embodiments of the application. It will be understood that each block of the flowchart and / or block diagram, and combinations of blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks.

[0208] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks.

[0209] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks. Figure 1 one or more functions specified in the flowchart and / or block diagram block or blocks.

[0210] While the preferred embodiments of the application have been described, additional variations and modifications can be made to the embodiments by those skilled in the art once they learn of the basic inventive concepts. Therefore, the appended claims are intended to cover all such modifications and variations as fall within the scope of the present application.

[0211] Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A method for measuring ocean topography by multi-source heterogeneous data fusion processing, characterized in that, The method comprises: Obtaining marine topographic data of different to-be-tested regions through multiple data sources to obtain multi-source heterogeneous data; Preprocessing the multi-source heterogeneous data to obtain preprocessed data; Extracting features from the preprocessed data and determining the weight of each data source based on the extracted features; According to the weight of each data source, the data of multiple data sources is fused to obtain the corresponding fusion result; According to the comprehensive evaluation index, the fusion result is evaluated, and the first fusion result that passes the evaluation is taken as the target marine topographic data; An iterative optimization function is constructed, and the second fusion result that fails the evaluation is iteratively optimized based on the iterative optimization function until the optimized second fusion result passes the evaluation, and the optimized second fusion result that passes the evaluation is taken as the target fusion result.

2. The method of claim 1, wherein, The method comprises: A first data of a deep water region is collected by a multi-beam sonar system carried by an unmanned boat; A second data of a shallow water region is collected by a hyperspectral camera carried by an unmanned aerial vehicle; A third data of a local high-precision is obtained by deploying an underwater laser scanner in a key area; The first data, the second data and the third data are obtained to obtain the multi-source heterogeneous data.

3. The method of claim 2, wherein, The method comprises: Noise in the multi-source heterogeneous data is filtered by an adaptive wavelet transform method to obtain first intermediate data; The first intermediate data is geometrically corrected and radiometrically corrected to obtain second intermediate data; The second intermediate data of different data sources are unified to the same coordinate system and resolution to obtain the preprocessed data.

4. The method of claim 3, wherein, The method comprises: Water depth data and water turbidity data collected by the data source are extracted from the preprocessed data; Data quality factors, first adjustment coefficients, second adjustment coefficients, third adjustment coefficients and fourth adjustment coefficients corresponding to the data source are obtained; The weight of the data source is determined according to the data quality factors, the first adjustment coefficients, the second adjustment coefficients, the third adjustment coefficients and the fourth adjustment coefficients corresponding to the data source, and the water depth data and the water turbidity data collected by the data source.

5. The method of claim 4, wherein, The method comprises: The observation value, the local topographic curvature and the slope of each data source are obtained; The observation value, the local topographic curvature and the slope of each data source are adjusted according to the first smoothing factor and the second smoothing factor to obtain the fusion result.

6. The method of claim 5, wherein, The fusion function is represented as: where F(x, y, z) is the result of the fusion function, U i is the observation value of the i-th data source, H is the local terrain curvature, S is the slope, and μ and ρ are the first and second smoothing factors, respectively.

7. The method of claim 6, wherein, The method comprises: The fusion result corresponding to each sampling point of each data source and the true value corresponding to the fusion result are obtained; The data redundancy of the fusion result is obtained; The terrain gradient in the fusion result is processed according to a first correction coefficient, and the data redundancy is processed according to a second correction coefficient, and a corresponding comprehensive evaluation index is determined by combining the fusion result and the true value; If the comprehensive evaluation index of the fusion result is greater than a preset threshold, the fusion result is determined as the first fusion result that passes the evaluation.

8. The method of claim 1, wherein, The comprehensive evaluation index is expressed as: Wherein, E is the comprehensive evaluation index, P i is the fusion result corresponding to the i th sampling point of each data source, T i is the true value of the fusion result of the i th sampling point, is the terrain gradient, R is the data redundancy, ∈ is the first correction coefficient, θ is the second correction coefficient, and n is the total number of sampling points.

9. The method of claim 8, wherein, The second fusion result that fails the evaluation is iteratively optimized based on the iterative optimization function of the comprehensive evaluation index, including: obtaining a preset learning rate, a convergence control parameter, and a Laplacian of an optimization variable of a current iteration round; obtaining a first fusion function value of the current iteration round and a second fusion function value of a previous iteration round of the current iteration round; iteratively optimizing the first fusion function value and the second fusion function value according to the learning rate, the convergence control parameter, and the Laplacian to obtain an optimized second fusion result.

10. A multi-source heterogeneous data fusion processing ocean topography measuring system for performing the multi-source heterogeneous data fusion processing ocean topography measuring method according to any one of claims 1 to 9, characterized by, The system comprises: a data acquisition module configured to acquire marine terrain data of different to-be-tested regions through a plurality of data sources to obtain multi-source heterogeneous data; a data processing module configured to preprocess the multi-source heterogeneous data to obtain preprocessed data; a weight determination module configured to extract features from the preprocessed data and determine a weight of each data source based on the extracted features; a data fusion module configured to fuse data of the plurality of data sources according to the weight of each data source to obtain a corresponding fusion result; a first processing module configured to evaluate the fusion result according to a comprehensive evaluation index and take a first fusion result that passes the evaluation as target marine terrain data; a second processing module configured to construct an iterative optimization function and iteratively optimize a second fusion result that fails the evaluation based on the iterative optimization function of the comprehensive evaluation index until the optimized second fusion result passes the evaluation, and take the optimized second fusion result that passes the evaluation as target fusion result.

Images