Remote data transmission method and system based on deep sea detection sonar

Abstract

The invention relates to the technical field of data transmission, in particular to a remote data transmission method and system based on deep-sea detection sonar, and the method comprises the steps: obtaining an original sonar image, and carrying out the preprocessing of the original sonar image, and obtaining a processing image in a coding block form; pixel point edge feature importance is quantized based on gradient magnitude and neighborhood gradient direction consistency, and coding block region importance is quantized in combination with gray level co-occurrence matrix texture features; calculating the signal-to-noise ratio reliability weight of the coding block, constructing an objective function containing an importance consistency item and a spatial continuity penalty item, minimizing the function to obtain optimized importance distribution of the coding block, and quantizing compression parameters; and compressing each coding block according to the compression parameters to complete remote transmission. According to the method, the sonar image is partitioned, and the compression parameters are adaptively allocated based on feature importance, so that the data volume is reduced on the premise of ensuring that the key features of the image are reserved, and the remote transmission efficiency and reliability of the deep-sea detection sonar image are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data transmission. In particular, it relates to a remote data transmission method and system based on a deep-sea exploration sonar. BACKGROUND

[0002] Deep-sea exploration is a core means for exploring marine resources, studying seabed geological structure, and monitoring marine environment. Sonar images are a key carrier for obtaining seabed target and environment information in deep-sea exploration. The efficiency and reliability of remote data transmission of sonar images directly determine the analysis value and application effect of deep-sea exploration data. Efficient sonar image remote transmission technology can realize real-time return and analysis of seabed exploration data, and provide accurate data support for decision-making and process control of deep-sea exploration operations. It is an indispensable part of the deep-sea exploration technology system.

[0003] In the existing deep-sea exploration sonar image transmission technology, the mainstream conventional processing method adopts a single strategy of filtering denoising combined with fixed proportion compression. This kind of method does not fully consider the importance difference of different regions of sonar images. It uses the same compression standard for key feature regions and non-key regions, which is difficult to achieve effective balance between improving data compression rate, reducing transmission data volume, preserving key features of sonar images, and ensuring data analysis accuracy. It is easy to cause problems such as low transmission efficiency due to insufficient compression rate or invalidation of exploration data due to feature loss.

[0004] Although some existing technologies try to use adaptive compression methods to optimize sonar image transmission effect, they only perform image blocking and importance judgment based on pixel gray value, without comprehensive analysis combining core features such as edges and textures of sonar images. This leads to serious loss of effective information of compressed sonar images, which cannot meet the actual needs of subsequent seabed information analysis. At the same time, this kind of method does not introduce signal-to-noise ratio reliability weight in the process of importance evaluation of coding blocks, which cannot effectively correct the signal quality of coding blocks under complex noise interference in deep sea, resulting in insufficient accuracy of coding block importance judgment, further reducing the overall effect of sonar image compression transmission, and making it difficult to adapt to the actual application scenario of deep-sea exploration. SUMMARY

[0005] To solve the problem that the existing adaptive compression method of sonar images only judges based on pixel gray value, does not fuse edge and texture features, and does not introduce signal-to-noise ratio reliability weight, resulting in loss of effective information of images, inaccurate importance judgment, and poor compression transmission effect, which is difficult to adapt to the actual needs of deep-sea exploration, the present application provides solutions in the following aspects.

[0006] In a first aspect, a deep-sea exploration sonar-based remote data transmission method comprises: obtaining a raw sonar image and preprocessing to obtain a processed image in the form of coding blocks; quantifying the edge feature importance of each pixel point in the processed image based on the gradient amplitude and the consistency of the gradient direction in the neighborhood; combining the edge feature importance with the texture features obtained based on the gray level co-occurrence matrix to quantify the importance of each coding block region in the processed image; calculating the signal-to-noise ratio reliability weight of each coding block, combining the signal-to-noise ratio reliability weight to construct a target function containing a coding block importance consistency term and a neighboring coding block importance difference penalty term, and solving the optimized coding block importance distribution by minimizing the target function to obtain the optimized coding block importance distribution; and quantifying the compression parameters of each coding block based on the optimized coding block importance distribution; and respectively compressing each coding block according to the corresponding compression parameters of each coding block, and remotely transmitting the compressed sonar image data.

[0007] Preferably, the calculation method of the edge feature importance comprises: Taking each pixel point of the preprocessed image as a target pixel point, calculating the gradient amplitude and gradient direction of the target pixel point, calculating the average cosine similarity of the gradient direction of the target pixel point and the pixel points in the neighborhood of the target pixel point, and calculating the edge feature importance of the corresponding pixel point by operating the normalized gradient amplitude and the average cosine similarity.

[0008] Preferably, the importance of each coding block region comprises: Multiplying the maximum value of the edge feature importance of all pixel points in each coding block with the ratio of the energy value of the corresponding coding block based on the gray level co-occurrence matrix to the energy value of the entire processed image based on the gray level co-occurrence matrix to obtain the importance of the corresponding coding block region.

[0009] Preferably, the step of calculating the signal-to-noise ratio reliability weight of each coding block comprises: Calculating the signal-to-noise ratio of each coding block and the average signal-to-noise ratio of all coding blocks, calculating the base weight by combining the signal-to-noise ratio of each coding block and the average signal-to-noise ratio of all coding blocks through an exponential function, and obtaining the signal-to-noise ratio reliability weight corresponding to each coding block by normalizing the base weight.

[0010] Preferably, the calculation method of the target function comprises: Taking any coding block as a target coding block, performing product operation on the signal-to-noise ratio reliability weight of the target coding block and the square of the importance difference before and after optimization of the corresponding target coding block, and summing the results to obtain the importance consistency. The difference weight of the target coding block corresponding to the adjacent coding block and the square of the importance difference value of the target coding block and the adjacent coding block after optimization are multiplied, the sum of the product results of all adjacent coding blocks is summed to obtain a single-group neighbor block difference penalty value, and the sum of the single-group neighbor block difference penalty values of all coding blocks is summed to obtain a spatial continuity penalty term. The sum of the importance consistency and the spatial continuity penalty term is taken as a target function of the target coding block, wherein the difference weight of each coding block and its adjacent coding block is obtained by normalizing the initial importance difference value of the two coding blocks and the standard deviation of the global coding block importance after exponential function operation.

[0011] Preferably, the compression parameters of each coding block are quantized based on the optimized coding block importance distribution, including: The upper and lower limit values of the quantization step length are determined, the maximum value and the minimum value of the optimized importance of all coding blocks are extracted, each coding block is traversed, the difference value between the global importance maximum value and the optimized importance of the current coding block is calculated, the ratio of the difference value to the difference between the global importance maximum value and the minimum value is calculated, the ratio is multiplied by the difference between the upper limit value and the lower limit value of the quantization step length, the multiplication result is rounded down, and then added to the lower limit value of the quantization step length to obtain the quantization step length of the current coding block, and the quantization step length is used as the compression parameter of each coding block.

[0012] Preferably, the step of obtaining the processed image in the form of coding blocks comprises: The original sonar image is denoised by using a wavelet denoising algorithm, and the denoised image is segmented into coding blocks of a preset size to obtain a processed image in the form of coding blocks.

[0013] In a second aspect, a remote data transmission system based on a deep-sea exploration sonar includes a processor and a memory, and the memory stores computer program instructions which, when executed by the processor, implement the remote data transmission method based on the deep-sea exploration sonar.

[0014] The present application has the following effects: 1、The present application quantifies the importance of the pixel edge features by fusing the gradient amplitude and the neighborhood gradient direction consistency, and quantifies the importance of the coding block region by combining the texture feature extracted by the gray level co-occurrence matrix, thereby breaking through the limitations of the prior art which only relies on the pixel gray value determination, accurately identifying the key feature region of the sonar image, greatly reducing the effective information loss of the compressed image, and ensuring the accuracy of subsequent seabed information analysis.

[0015] 2、The application can correct the coding block signal quality under the interference of complex deep-sea noise by introducing the signal-to-noise ratio reliability weight in the coding block importance evaluation, improve the accuracy of coding block importance judgment, and construct a target function containing an importance consistency term and a spatial continuity penalty term, so that the coding block importance distribution obtained after optimization has rationality and spatial continuity, and effectively improves the overall effect of sonar image compression transmission.

[0016] 3、The application realizes block-level adaptive compression of sonar images by using the optimized coding block importance distribution for differentiated quantization compression parameters of each coding block, greatly reduces the transmission data volume, improves the remote transmission efficiency of deep-sea exploration sonar images, adopts a low-loss compression strategy for important feature areas, and balances the compression rate and feature retention degree, perfectly adapting to the actual application requirements of deep-sea exploration. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 is a method flowchart of steps S1-S4 in a remote data transmission method based on a deep-sea exploration sonar according to an embodiment of the application.

[0018] Figure 2 is a structural block diagram of a remote data transmission system based on a deep-sea exploration sonar according to an embodiment of the application. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application.

[0020] Referring to Figure 1 A remote data transmission method based on a deep-sea exploration sonar includes steps S1-S4, and specifically as follows: S1: Obtain an original sonar image and perform preprocessing to obtain a processed image in the form of coding blocks.

[0021] The original sonar image of the seabed is collected by a deep-sea exploration sonar device, the original sonar image is imported into a processing terminal, a wavelet denoising algorithm is used to perform denoising processing on the original sonar image, db4 wavelet is selected as the base wavelet, the image is decomposed into 3 layers of wavelets, and the image is reconstructed after removing the high-frequency noise components; the denoised image is uniformly segmented according to the preset size of 64x64 to obtain a plurality of coding blocks, and a processed image in the form of coding blocks is formed, and if the edge size of the image is less than 64x64, the zero padding method is used to complete the preset size.

[0022] S2: Quantify the edge feature importance of each pixel point in the processed image based on the gradient amplitude and the consistency of the neighborhood gradient direction, and combine the edge feature importance with the texture feature obtained based on the gray level co-occurrence matrix to quantify the importance of each coding block region in the processed image.

[0023] The Sobel gradient operator is used to calculate the gradient amplitude and gradient direction of each pixel point in the processed image. Taking any pixel point as a target pixel point, a neighborhood range of 3x3 is selected, the cosine similarity of the gradient direction of the target pixel point and the 8 pixel points in the neighborhood is calculated, and the mean value of the 8 cosine similarities is calculated. The gradient amplitude of the target pixel point is normalized by maximum and minimum value, and the normalized gradient amplitude is multiplied by the mean value of the cosine similarity to obtain the edge feature importance of the target pixel point. All pixel points are traversed to complete the quantification of the edge feature importance of all pixel points.

[0024] All coding blocks in the processed image are traversed. The edge feature importance data of all pixel points in the coding block is extracted, and the maximum value is selected, which represents the key degree of the edge feature of the corresponding coding block. The energy value of the coding block and the energy value of the entire processed image are calculated based on the gray level co-occurrence matrix. The ratio of the energy value of the coding block to the energy value of the entire processed image is calculated, which represents the proportion of the texture feature of the corresponding coding block in the entire image. The maximum value of the edge feature importance of the coding block and the proportion of the texture feature are multiplied to obtain the importance quantization value of the corresponding coding block region. The importance of all coding block regions is quantified, and the comprehensive importance of the edge and texture features of the coding block is determined.

[0025] S3: Calculate the signal-to-noise ratio reliability weight of each coding block, combine the signal-to-noise ratio reliability weight to construct a target function containing a coding block importance consistency term and a neighboring coding block importance difference penalty term, and solve the optimized coding block importance distribution by minimizing the target function. Based on the optimized coding block importance distribution, the compression parameters of each coding block are quantified.

[0026] All coding blocks in the processed image are traversed, and the signal-to-noise ratio of each coding block is calculated. The signal-to-noise ratio reflects the relative relationship between the signal quality and noise interference of the coding block. The higher the signal-to-noise ratio value, the higher the proportion of effective signal and the smaller the noise interference.

[0027] The global signal-to-noise ratio mean value of all coding block signal-to-noise ratio values is calculated. The global signal-to-noise ratio mean value is used as the overall reference benchmark for the signal quality of the coding block.

[0028] By introducing an exponential function, the signal-to-noise ratio (SNR) value of each coding block and the global SNR mean are substituted into the exponential function to complete the calculation, resulting in the basic weight corresponding to each coding block. The operational characteristics of the exponential function enable the differentiation of weights for coding blocks with different SNRs. Coding blocks with high SNR are matched with higher basic weights, highlighting the signal reliability of high SNR coding blocks.

[0029] The basic weights of all coding blocks are normalized to eliminate the dimensional differences in the weight values, so that the signal-to-noise ratio reliability weights of all coding blocks are in a uniform range. This results in the signal-to-noise ratio reliability weight for each coding block, which serves as the basis for signal quality correction in the subsequent coding block importance optimization process.

[0030] Specifically, the signal-to-noise ratio reliability weight satisfies the following relationship: ; In the formula, Indicates the first The signal-to-noise ratio reliability weight of each coding block Indicates the first Signal-to-noise ratio of each coded block Represents the mean signal-to-noise ratio of the global coded block, used as a threshold parameter when... When the signal-to-noise ratio (SNR) reliability weight is high, it is close to 1, indicating that the importance estimate is reliable; when When the signal-to-noise ratio reliability weight is low, it indicates that the importance estimate of this block needs to be adjusted. It is a normalization function that makes the sum of the weights equal to 1. This indicates normalization processing. Represented by natural constant An exponential function with base 0.

[0031] Select any coded block in the image to be processed as the target coded block and perform the objective function construction operation. Calculate the product of the signal-to-noise ratio reliability weight of each coded block and the square of the difference in importance before and after optimization. Perform a summation operation on the product results of all coded blocks. The result of this summation operation is defined as the importance consistency term. The importance consistency term is used to constrain the deviation between the importance of the coded block after optimization and the initial importance, ensuring the rationality of the coded block importance optimization results.

[0032] Calculate the difference weight between each coding block and its corresponding neighboring coding blocks. Calculate the product of this difference weight and the square of the importance difference between the corresponding coding block and its neighboring coding blocks after optimization. Perform a summation operation on the product results of each coding block and all neighboring coding blocks. The result of this summation operation is defined as a single-group neighboring block difference penalty value. Perform a summation operation on the single-group neighboring block difference penalty values ​​of all coding blocks in the processed image. The result of this summation operation is defined as a spatial continuity penalty term. The spatial continuity penalty term is used to penalize the importance difference between neighboring coding blocks, avoiding image distortion caused by excessive importance deviation after optimization of neighboring coding blocks.

[0033] The sum of the importance consistency term and the spatial continuity penalty term is calculated, and this sum is used as the objective function for the processed image. The objective function provides a basis for the optimization of the importance distribution of the coding block.

[0034] The difference weights between each coding block and its adjacent coding blocks are obtained as follows: the initial importance difference between the two coding blocks is calculated, the initial importance difference and the standard deviation of the global coding block importance are substituted into the exponential function to complete the calculation, and the result of the exponential function calculation is normalized. The result after normalization is the difference weight between the two corresponding coding blocks.

[0035] Specifically, the objective function satisfies the following polynomial: ; ; In the formula, This represents the objective function for optimizing the importance of coded blocks. Indicates the first The signal-to-noise ratio reliability weight of each coding block Indicates the first The optimized importance quantization value of each coding block Indicates the first The initial importance quantization value of each coded block, Indicates the first The set of adjacent coded blocks of a coded block, Indicates the first The coded block and the Difference weights between coded blocks Indicates the first The optimized importance quantization value of each coding block Indicates the first Initial importance quantization value of each coding block, The standard deviation representing the importance of global coding blocks, This indicates normalization processing. Represented by natural constant An exponential function with base 0.

[0036] With objective function To optimize the core decision-making criteria, a genetic algorithm is used to optimize the objective function. By performing a minimization operation and leveraging the selection, crossover, and mutation iterative optimization features of a genetic algorithm, the importance quantization value of the code block is optimized through multiple generations of evolution, gradually converging to the objective function. The minimum solution corresponds to the set of importance quantization values ​​of all coded blocks after optimization. This set is the importance distribution of the coded blocks after optimization. Based on this importance distribution, the spatial consistency optimization of the importance of the coded blocks is completed, and finally the importance of the coded blocks after spatial consistency optimization is obtained, realizing the dual optimality of the rationality and spatial continuity of the importance of the coded blocks in the global scope.

[0037] Pre-set upper and lower limits for the quantization step size as numerical boundary benchmarks for quantization step size allocation. Extract the maximum and minimum values ​​of the importance quantization values ​​of all coding blocks after spatial consistency optimization, and use them as global reference extreme values ​​for the importance of the optimized coding blocks.

[0038] The process iterates through all coded blocks in the image, calculating the difference between the global maximum importance value of the optimized coded block and the current coded block's optimized importance quantization value. This difference reflects the degree of deviation of the current coded block's importance from the global highest importance. The ratio of this deviation difference to the difference between the global maximum and global minimum importance values ​​of the optimized coded block is then calculated. This ratio is used to normalize the current coded block's importance, eliminating dimensional differences in the importance quantization values.

[0039] The normalized ratio is multiplied by the difference between the upper and lower limits of the quantization step size to achieve a linear mapping of the numerical range of the vectorized step size for the importance of the coding block. The result of the linear mapping is then rounded down to an integer to match the parameter value requirements of the compression algorithm.

[0040] The result after rounding down is added to the lower limit of the quantization step size to obtain the quantization step size corresponding to the current coding block. This quantization step size directly represents the compression degree of the corresponding coding block.

[0041] The quantization step size corresponding to each coding block is used as the core parameter for compression processing of each coding block, and differentiated compression strategies are assigned to coding blocks of different importance based on this parameter.

[0042] Specifically, the compression parameters of each coded block satisfy the following relationship: ; In the formula, Indicates the first quantization step size of each coding block, importance of coding blocks The lower the value, the larger the quantization step size, resulting in lower similarity between the compressed image and the original image, and greater compression loss; conversely, the higher the value, the larger the quantization step size. The smaller the value, the smaller the compression loss. These represent the upper and lower limits of the quantization step size, respectively, and are set to 10 and 1. The values ​​can be selected based on actual conditions and experience. Indicates the first The importance quantization value of each coded block after spatial consistency optimization. This represents the maximum importance quantization value among all coded blocks after spatial consistency optimization. This represents the minimum importance quantization value among all coded blocks after spatial consistency optimization. This indicates rounding down to the nearest integer.

[0043] S4: Compress each coding block according to the compression parameters corresponding to each coding block, and then remotely transmit the compressed sonar image data.

[0044] Choose the JPEG algorithm or other image compression algorithms that support block-level quantization step size configuration. These algorithms can independently configure compression parameters at the coding block level, adapting to the needs of differentiated compression of coding blocks.

[0045] The calculated quantization step size parameters for each coding block are matched to the corresponding coding blocks in the processed image. The quantization step size parameters are embedded into the compression operation process of the corresponding coding blocks. Based on the numerical differences of the quantization step size parameters, differentiated compression processing is performed for each coding block. Coding blocks with high importance are matched with small quantization step sizes to achieve low-loss compression, while coding blocks with low importance are matched with large quantization step sizes to achieve high-ratio compression.

[0046] After the compression operation of all coded blocks in the image to be processed is completed, a data reassembly operation is performed on all compressed coded blocks to form complete compressed sonar image data. The compressed sonar image data is then sent to the receiving terminal through a preset transmission link to complete the remote transmission of deep-sea exploration sonar images.

[0047] This invention also provides a remote data transmission system based on deep-sea sonar. For example... Figure 2 As shown, the system includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a remote data transmission method based on deep-sea sonar according to the first aspect of the present invention. The system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface. Their configuration and functions are known in the art and will not be described further here.

[0048] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for long-range data transmission based on deep-sea sonar, characterized in that, include: The raw sonar image is acquired and preprocessed to obtain a processed image in the form of coded blocks; Based on the gradient magnitude and the consistency of the gradient direction in the neighborhood, the importance of edge features of each pixel in the processed image is quantified. Combining the importance of edge features with the texture features obtained based on the gray-level co-occurrence matrix, the importance of each coded block region in the processed image is quantified. Calculate the signal-to-noise ratio reliability weight of each coding block, and construct an objective function that includes a coding block importance consistency term and an adjacent coding block importance difference penalty term by combining the signal-to-noise ratio reliability weight. Solve the objective function to obtain the optimized coding block importance distribution, and quantify the compression parameters of each coding block based on the optimized coding block importance distribution. Each coding block is compressed according to the compression parameters corresponding to each coding block, and the compressed sonar image data is transmitted remotely.

2. The remote data transmission method based on deep-sea sonar according to claim 1, characterized in that, The methods for calculating the importance of the edge features include: Taking each pixel of the preprocessed image as the target pixel, calculate the gradient magnitude and gradient direction of the target pixel, calculate the mean cosine similarity of the gradient direction between the target pixel and the pixels in its neighborhood, and then calculate the edge feature importance of the corresponding pixel by normalizing the gradient magnitude and calculating the mean cosine similarity.

3. The method for remote data transmission based on deep-sea sonar according to claim 1, characterized in that, The importance of each coded block region includes: The maximum value of the edge feature importance of all pixels in each coding block is obtained. The maximum value is then multiplied by the ratio of the energy value of the corresponding coding block based on the gray-level co-occurrence matrix to the energy value of the entire processed image based on the gray-level co-occurrence matrix to obtain the importance of the corresponding coding block region.

4. The method for remote data transmission based on deep-sea sonar according to claim 1, characterized in that, The steps for calculating the signal-to-noise ratio reliability weight of each coded block include: The signal-to-noise ratio (SNR) of each coding block and the mean SNR of all coding blocks are calculated. The basic weights are calculated by combining the SNR of each coding block and the mean SNR of all coding blocks using an exponential function. The basic weights are then normalized to obtain the SNR reliability weights corresponding to each coding block.

5. The method for remote data transmission based on deep-sea sonar according to claim 1, characterized in that, The objective function is calculated in the following ways: Taking any coding block as the target coding block, the signal-to-noise ratio reliability weight of the target coding block is multiplied by the square of the difference in importance before and after optimization of the corresponding target coding block, and the sum is taken as the importance consistency. The difference weights of the target coding block corresponding to the adjacent coding blocks are multiplied by the square of the difference in importance between the target coding block and the adjacent coding blocks after optimization. The product results of all adjacent coding blocks are summed to obtain a single set of neighboring block difference penalty values. The single set of neighboring block difference penalty values ​​of all coding blocks are summed to obtain the spatial continuity penalty term. The sum of the importance consistency and spatial continuity penalty terms is used as the objective function of the target coding block; where the difference weight between each coding block and its neighboring coding blocks is obtained by normalizing the initial importance difference between the two coding blocks and the standard deviation of the global coding block importance after performing an exponential function operation.

6. The method for remote data transmission based on deep-sea sonar according to claim 1, characterized in that, The compression parameters of each coding block are quantized based on the optimized coding block importance distribution, including: Determine the upper and lower limits of the quantization step size, extract the maximum and minimum values ​​of the importance of all coded blocks after optimization, traverse each coded block, calculate the difference between the global maximum importance and the current coded block's optimized importance, calculate the ratio of this difference to the difference between the global maximum and minimum importance, multiply this ratio by the difference between the upper and lower limits of the quantization step size, round the result down, and then add it to the lower limit of the quantization step size to obtain the quantization step size of the current coded block. Use the quantization step size as the compression parameter for each coded block.

7. The method for remote data transmission based on deep-sea sonar according to claim 1, characterized in that, The steps for obtaining the processed image in the form of the encoded blocks include: The original sonar image is denoised using a wavelet denoising algorithm. The denoised image is then divided into coded blocks of a preset size to obtain a processed image in the form of coded blocks.

8. A remote data transmission system based on deep-sea sonar, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement the remote data transmission method based on deep-sea sonar according to any one of claims 1-7.

Images