Self-adaptive target identification method of single-beam mechanical scanning type sonar system

Abstract

The invention discloses a self-adaptive target identification method of a single-beam mechanical scanning type sonar system, which has the advantages that an identification strategy from coarse scanning to fine scanning is adopted, wide-area wide-beam scanning is carried out in an initial stage, and then focusing is carried out in a potential target area; therefore, the precision of a short-distance target is improved while the long-distance detection capability is maintained; by dynamically adjusting scanning parameters, such as automatically switching the detection range and adjusting the scanning angle, the balance between the detection precision and the imaging distance can be remarkably improved, the scanning time is remarkably shortened, the response speed of the system is improved, and the system is more suitable for a real-time detection task in a dynamic environment; the learning model is introduced for adaptive target identification, and the system can optimize a data processing flow, can process scanning data more efficiently and reduce redundant information, so that the imaging definition can be remarkably improved while the burden of data processing is reduced, and particularly the detail presentation of a key area can be remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of acoustic imaging detection, and in particular to an adaptive target recognition method for a single-beam mechanical scanning sonar system. Background Art

[0002] In current sonar systems, detection accuracy and imaging range often present an irreconcilable conflict. Generally speaking, improving detection accuracy reduces imaging range, while pursuing longer imaging ranges sacrifices accuracy. This limitation significantly restricts the applicability of sonar systems, especially in applications requiring both long-range detection and high-precision identification.

[0003] Another common problem with existing sonar systems is their relatively insufficient data processing capabilities, which directly leads to low image clarity. This issue is particularly prominent in applications requiring high-resolution images, such as underwater target identification or environmental surveys, limiting the effectiveness of sonar systems. Because mechanical scanning sonar relies on a point-by-point scanning process, it is slow and cannot meet real-time requirements. This becomes a major bottleneck in dynamic environments or scenarios requiring rapid response. Existing single-beam devices lack dynamic adjustment capabilities, and no single-beam device currently on the market offers this capability, making development urgent. Summary of the Invention

[0004] A technical problem to be solved by this application is to overcome the defects of the above-mentioned related technologies and provide an adaptive target recognition method for a single-beam mechanical scanning sonar system. Through an adaptive control module, the scanning parameters are dynamically adjusted based on real-time analysis of echo data to achieve "coarse to fine" target recognition.

[0005] The technical solution adopted by the present invention to solve the technical problem is: an adaptive target recognition method for a single-beam mechanical scanning sonar system, which specifically includes the following steps: S1 coarse scan: The single-beam mechanical scanning sonar performs a 360° wide-beam scan at the first step angle to quickly acquire echo data; it also performs preliminary feature extraction on the echo data to identify potential target areas; S2 Target Detection and Marking: Analyze the extracted preliminary features to determine and mark potential target areas; Detect the frequency offset of the echo data through Doppler effect analysis. If the target is determined to be dynamic, proceed to step S3A; if the target is determined to be static, proceed to step S3B; S3A adaptive range adjustment: Use the decision tree algorithm to determine whether the potential target area is at the range boundary. Adjust to the appropriate range based on the distance characteristics of the potential target area and the preset boundary threshold, and then enter step S4A; S4A fine scanning: Perform local narrow beam scanning on the marked potential target area using an appropriate range and a second step angle to obtain high-resolution target detail data, and then proceed to step S5; the second step angle is half of the first step angle; S3B mid-range scanning and adaptive range adjustment: Perform mid-range confirmation scanning on the potential target area, calibrate the angle range and distance characteristics of the target area, use the decision tree algorithm to determine whether the calibrated target area is at the range boundary, adjust to the appropriate range, and enter step S4B; S4B fine scanning: Perform narrow beam scanning on the calibration target area using an appropriate range and a third step angle to obtain high-resolution target detail data, and then proceed to step S5; wherein the third step angle is 1% of the angular range of the calibration target area; S5 Target Recognition: Use the learning model to perform target recognition on the target detail data obtained by the fine scanning in step S4A or step S4B, and output the target category and location information.

[0006] As a preference, it also includes S6 parameter optimization and adaptive learning: using learning algorithms to automatically adjust scanning parameters and hyperparameters of the learning model based on historical recognition results and system performance feedback.

[0007] Compared with related technologies, the advantage of the present invention lies in the adoption of a coarse-to-fine scanning recognition strategy. The single-beam mechanical scanning sonar system can perform a wide-area scan in the initial stage and then focus on the potential target area, thereby improving the accuracy of close-range targets while maintaining long-range detection capabilities. Depending on the state of the target, it can directly switch from coarse scanning to fine scanning for high-speed target tracking, or add a medium-range confirmation scan between coarse scanning and fine scanning. It is mainly used for tasks such as underwater archaeology that require large-scale coverage and detailed identification. By dynamically adjusting scanning parameters, such as automatically switching the detection range and adjusting the scanning angle, the balance between detection accuracy and imaging distance can be significantly improved, while significantly shortening the scanning time and improving the response speed of the single-beam mechanical scanning sonar system, making it more suitable for real-time detection tasks in dynamic environments. By introducing a learning model for adaptive target recognition, the single-beam mechanical scanning sonar system can optimize the data processing process, process scanning data more efficiently and reduce redundant information, thereby reducing the burden of data processing while significantly improving imaging clarity, especially in the presentation of details in key areas.

[0008] Preferably, the learning algorithm in step S6 is a reinforcement learning algorithm, and the scanning parameters include beam width, step angle, and fine scanning range. The reinforcement learning (RL) algorithm enables the single-beam mechanical scanning sonar system to continuously optimize the recognition strategy and adapt to different environments and targets.

[0009] Preferably, in step S1, a convolutional neural network is used to extract preliminary features from the echo data, and the preliminary features include intensity and texture.

[0010] Preferably, in step S2, a support vector machine classifier is used to analyze the preliminary features extracted by the convolutional neural network. The combination of the convolutional neural network (CNN) and the support vector machine (SVM) significantly improves the accuracy and efficiency of data processing and reduces misjudgments and missed judgments.

[0011] Preferably, the learning model in step S5 is a recurrent neural network, and the target detail data is time-series echo data. Recurrent neural networks can handle temporal dependencies and accurately identify the dynamic characteristics of targets. By analyzing temporal dependencies, the accuracy and efficiency of target recognition can be significantly improved. This single-beam mechanical scanning sonar system fuses sonar and optical sensor data to enhance system robustness through collaborative analysis of multi-dimensional information. Data fusion utilizes feature-level fusion, extracting features from sonar and optical data and inputting them into a joint model (which can be analyzed using a deep learning model or convolutional neural network). This multi-source fusion significantly reduces false positive and false negative rates, ultimately resulting in a comprehensive recognition result and a clear target image.

[0012] As an improvement, during the adaptive range adjustment in step S3A or step S3B, if the target is at the boundary, the range is increased; if the target is not at the boundary and has sufficient detail, the range is decreased. Increasing the range to cover a wider area, while decreasing the range to improve accuracy, improves the balance between detection accuracy and imaging distance.

[0013] Preferably, the scanning angle of the local narrow beam in step S4A is 180° or 120° of the potential target area containing the mark.

[0014] Preferably, the first step angle is 3.6° and the second step angle is 1.8°. 360° can be divided into integer parts, which facilitates subsequent model calculations and algorithm calculations. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of the adaptive target recognition method of the present invention. DETAILED DESCRIPTION

[0016] First, those skilled in the art should understand that these embodiments are merely used to explain the technical principles of the embodiments of the present application and are not intended to limit the scope of protection of the embodiments of the present application. Those skilled in the art may adjust them as needed to suit specific application scenarios.

[0017] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] This embodiment provides an adaptive target recognition method for a single-beam mechanical scanning sonar system, which specifically includes the following steps: S1. Coarse scan: The single-beam mechanical scanning sonar performs a 360° wide beam scan with a 3.6° step angle to quickly acquire echo data. Single-beam mechanical scanning sonar is also called single-beam mechanical scanning sonar or single-beam mechanical scanning sonar.

[0019] A convolutional neural network (CNN) is used to perform preliminary feature extraction on echo data and identify potential target areas. The CNN model is trained to learn the initial characteristics of targets in echo signals and automatically extract advanced features such as intensity and texture, replacing traditional threshold judgment.

[0020] S2. Object detection and labeling: A support vector machine (SVM) classifier analyzes the features extracted by the CNN to identify and mark potential target areas. Through supervised learning, the SVM classifier effectively distinguishes targets from noise, improving target detection accuracy. Furthermore, the location information of the marked potential target areas is stored, occupying minimal memory (less than half a byte). Doppler effect analysis is used to detect the frequency offset of the echo data. If the target is dynamic, step S3A is entered. This means that in high-speed target tracking tasks, moving targets cause significant frequency shifts, prompting step S3A.

[0021] S3A. Range adaptive adjustment: A decision tree algorithm is used to determine whether a target is at the range boundary. The decision tree automatically determines whether to extend or shorten the range based on the target's distance characteristics and a preset boundary threshold.

[0022] If the target is located at the edge, the single-beam mechanical scanning sonar system automatically increases its range to cover a wider area. If the target is not located at the edge and has sufficient detail, the single-beam mechanical scanning sonar system decreases its range to improve accuracy. The process proceeds to step S4A. The distance characteristics of the potential target area are within 1 / 3 to 2 / 3 of the appropriate range.

[0023] S4A. Detailed scan: The marked target area is scanned with a local narrow beam of 180° or 120° at a step angle of 1.8°. Compared with the traditional 360° full-range fine scanning, the scanning speed is increased by 4 times or 6 times, and high-resolution target detail data is obtained.

[0024] S5. Target Identification: A recurrent neural network (RNN) is used to identify targets using time-series echo data from fine-scan scans. The RNN model can handle temporal dependencies and accurately identify dynamic characteristics of targets. The model outputs target category and location information, achieving high-precision target recognition.

[0025] S6. Parameter Optimization and Adaptive Learning: A reinforcement learning (RL) algorithm is introduced to automatically adjust scanning parameters (such as step angle and scanning range) and model hyperparameters based on historical recognition results and system performance feedback. This RL algorithm enables this single-beam mechanical scanning sonar system to continuously optimize its recognition strategy and adapt to different environments and targets.

[0026] Here, wide beam refers to a narrowband beam with a larger fan angle, and narrow beam refers to a narrowband beam with a smaller fan angle. That is, the fan angle of the wide beam is larger than the fan angle of the narrow beam. Usually, the fan angle of the wide beam is 1.5 to 3 times that of the narrow beam.

[0027] By dynamically adjusting scanning parameters (such as automatically switching the detection range and adjusting the scanning angle), the balance between detection accuracy and imaging range can be significantly improved. Using a "coarse-to-fine" recognition strategy, this single-beam mechanical scanning sonar system initially scans a wide area before focusing on areas of interest (potential target areas). This approach improves accuracy for close-range targets while maintaining long-range detection capabilities. This approach represents a significant breakthrough in existing technologies. By introducing an adaptive target recognition method, the proposed system optimizes the data processing pipeline, enabling more efficient processing of scan data and reducing redundant information. This approach not only reduces the data processing burden but also significantly improves imaging clarity, especially in detail rendering in critical areas. Although the real-time performance of mechanical scanning sonar is limited by its physical characteristics, the proposed adaptive algorithm effectively improves the response speed and operational efficiency of this single-beam mechanical scanning sonar system by automatically adjusting and optimizing the scanning process (for example, prioritizing scanning target areas). This improvement offers significant real-time advantages over traditional systems. By combining mechanical scanning sonar with adaptive recognition methods, the proposed single-beam mechanical scanning sonar system can adjust scanning parameters in real time according to the target environment. This feature is unique in current technologies and opens up new possibilities for the application of single-beam sonar.

[0028] The "coarse-to-fine" step-by-step scanning strategy effectively balances accuracy and range. Coarse scanning quickly covers a large area, while fine scanning enhances detail recognition, enabling this single-beam mechanical scanning sonar system to detect long-range targets while maintaining high-resolution imaging quality. The combination of CNN and SVM significantly improves the accuracy and efficiency of data processing, reducing false positives and missed detections. Ultimately, image clarity is significantly improved, and target recognition accuracy is also enhanced, providing users with more reliable detection results. Compared with existing technologies, the synergistic effect of CNN and SVM increases image clarity from 80% to 95%. Compared with traditional 360° full-range fine scanning, scanning speed is increased by 4 times (180° area) or 6 times (120° area), respectively. This significantly shortens scanning time and improves the response speed of this single-beam mechanical scanning sonar system, making it more suitable for real-time detection tasks in dynamic environments. This single-beam mechanical scanning sonar system dynamically adjusts its operating parameters based on real-time data, maintaining optimal performance in complex or changing environments. This adaptive capability enhances its stability and practicality, making it particularly suitable for dynamic scenarios such as underwater detection. By analyzing temporal dependencies, RNN significantly improves the accuracy and efficiency of target recognition. This system excels in identifying dynamic targets, such as moving objects, providing users with more precise detection results.

[0029] In step S2, the frequency offset of the echo data is detected through Doppler effect analysis. If the target is determined to be static, step S3B is entered. That is, when it is mainly used for tasks such as underwater archaeology that require large-scale coverage and detail identification, the static target has no obvious offset and step S3B is entered.

[0030] S3B: Mid-range scan and adaptive range adjustment: A mid-range confirmation scan is performed on the potential target area. The target area's angular range and distance characteristics are calibrated. A decision tree algorithm is used to determine whether the calibrated target area is at the range boundary. The range is adjusted to the appropriate range, and the process proceeds to step S4B. During the mid-range confirmation scan, the range used is twice the maximum value of the potential target area's distance characteristics. The distance characteristics of the calibrated target area are within 1 / 3 to 2 / 3 of the appropriate range.

[0031] S4B fine scanning: perform narrow beam scanning on the calibration target area using an appropriate range at a third step angle to obtain high-resolution target detail data, and proceed to step S5; wherein the third step angle is one percent of the angle range of the calibration target area.

[0032] The mid-range confirmation scan focuses on the area detected by the wide-area scan (i.e., coarse scan) to verify whether it is a real target and eliminate false positives. The fine scan then performs a high-resolution scan on the confirmed target to obtain detailed feature information.

[0033] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. An adaptive target recognition method for a single-beam mechanical scanning sonar system, characterized in that: The specific steps include: S1 coarse scan: The single-beam mechanical scanning sonar performs a 360° wide-beam scan at the first step angle to quickly acquire echo data; it also performs preliminary feature extraction on the echo data to identify potential target areas; S2 Target Detection and Marking: Analyze the extracted preliminary features, determine and mark potential target areas; By analyzing the Doppler effect, the frequency offset of the echo data is detected. If the target is determined to be dynamic, the process proceeds to step S3A. If the target is determined to be static, the process proceeds to step S3B. S3A adaptive range adjustment: Use the decision tree algorithm to determine whether the potential target area is at the range boundary. Adjust to the appropriate range based on the distance characteristics of the potential target area and the preset boundary threshold, and then enter step S4A; S4A fine scanning: Perform local narrow beam scanning on the marked potential target area using an appropriate range and a second step angle to obtain high-resolution target detail data, and then proceed to step S5; the second step angle is half of the first step angle; S3B mid-range scanning and adaptive range adjustment: Perform mid-range confirmation scanning on the potential target area, calibrate the angle range and distance characteristics of the target area, use the decision tree algorithm to determine whether the calibrated target area is at the range boundary, adjust to the appropriate range, and enter step S4B; S4B fine scanning: Perform narrow beam scanning on the calibration target area using an appropriate range and a third step angle to obtain high-resolution target detail data, and then proceed to step S5; wherein the third step angle is 1% of the angular range of the calibration target area; S5 Target Recognition: Use the learning model to perform target recognition on the target detail data obtained by the fine scanning in step S4A or step S4B, and output the target category and location information.

2. The adaptive target recognition method of a single-beam mechanical scanning sonar system according to claim 1, characterized in that: It also includes S6 parameter optimization and adaptive learning: using learning algorithms to automatically adjust scanning parameters and hyperparameters of the learning model based on historical recognition results and system performance feedback.

3. The adaptive target recognition method for a single-beam mechanical scanning sonar system according to claim 2, characterized in that: The learning algorithm in step S6 is a reinforcement learning algorithm, and the scanning parameters include beam width, step angle, and fine scanning range.

4. The adaptive target recognition method for a single-beam mechanical scanning sonar system according to any one of claims 1 to 3, characterized in that: In step S1, a convolutional neural network is used to extract preliminary features of the echo data, and the preliminary features include intensity and texture.

5. The adaptive target recognition method for a single-beam mechanical scanning sonar system according to claim 4, characterized in that: In step S2, a support vector machine classifier is used to analyze the preliminary features extracted by the convolutional neural network.

6. The adaptive target recognition method for a single-beam mechanical scanning sonar system according to claim 5, characterized in that: The learning model of step S5 is a recurrent neural network, and the target detail data is time series echo data.

7. The adaptive target recognition method for a single-beam mechanical scanning sonar system according to any one of claims 1 to 3, characterized in that: In the adaptive range adjustment of step S3A or step S3B, if the target is at the boundary, the range is expanded; if the target is not at the boundary and the details are sufficient, the range is shortened.

8. The adaptive target recognition method for a single-beam mechanical scanning sonar system according to any one of claims 1 to 3, characterized in that: In step S4A, the scanning angle of the local narrow beam is 180° or 120° of the potential target area containing the mark.

9. The adaptive target recognition method for a single-beam mechanical scanning sonar system according to claim 8, characterized in that: The first step angle is 3.6° and the second step angle is 1.8°.

Images