Collision early warning method, system and equipment based on target detection and identification information

Abstract

The invention relates to the field of ship collision prevention, and provides a collision early warning method, system and device based on target detection and identification information, and the method comprises the steps: obtaining a to-be-processed image through the shooting of a camera, receiving the AIS information of other ships through a receiving device, and obtaining a first target position; determining data processing parameters and sea area environment parameters, calculating total processing delay, and performing signal filtering on the to-be-processed image to obtain a filtered image; performing parameter configuration on a target detection algorithm, obtaining interference characteristics through the filtered image, training a weight matrix, and obtaining a second target position through the interference characteristics and the weight matrix by the target detection algorithm; comprehensive similarity scores are calculated for screening, and potential high-risk targets are obtained; and performing position verification to obtain a high-risk target, obtaining a dynamic safety distance according to the wave height, calculating a collision risk index through the dynamic safety distance and the total processing delay, and performing collision early warning according to the collision risk index. According to the invention, the accuracy of collision early warning can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of ship anti-collision, and particularly to a collision warning method, system and device based on target detection and recognition information. Background Art

[0002] During the patrol process, patrol ships often confront and drive away ships attempting to invade the patrol area. Inevitably, collisions and conflicts will occur with the other party, which requires a reliable collision warning system. The current maritime collision warning system mainly faces various technical bottlenecks. The traditional warning method relying on a single sensor has obvious defects: the performance of the radar system drops sharply under bad sea conditions and it is difficult to stably track small targets; relying on AIS (Automatic Identification System) has the situation that ships attempting to invade will turn off the AIS or forge AIS signals. Although visual detection technology can provide auxiliary information, it is limited by the complex lighting conditions and dynamic wave interference on the sea surface, and a large number of misjudgments are likely to occur during the use of conventional algorithms. Existing systems are mostly designed for fixed scenarios such as ports, and insufficient consideration is given to the real-time performance and environmental adaptability required for maritime patrols. The data of various sensors are often processed in isolation, lacking an effective fusion mechanism, resulting in one-sided decision-making information. The existing solutions internationally generally have problems such as large response delays and weak small target recognition capabilities, and it is difficult to meet the safety requirements of patrol ships in high-mobility patrol scenarios. Especially when facing invading ships that deliberately turn off the AIS, traditional technical means seem inadequate. These technical shortcomings pose potential safety hazards during the maritime patrol process, and there is an urgent need to develop a new generation of intelligent multi-source fusion warning systems. Summary of the Invention

[0003] The present invention aims to solve at least one of the technical problems existing in the related art. For this purpose, the present invention provides a collision warning method, system and device based on target detection and recognition information, which realizes timely warning of potential collision risks by combining AIS information and image recognition results.

[0004] The present invention provides a collision warning method, system and device based on target detection and recognition information, including: S1: Deploy a camera and a receiving device on the target ship. The camera captures an image to be processed, and the receiving device receives the AIS information of other ships, thereby obtaining the first target position; S2: Determine data processing parameters and sea area environment parameters, calculate the total processing delay according to the data processing parameters, and perform signal filtering on the image to be processed according to the sea area environment parameters to obtain a filtered image; S3: Obtain the target detection algorithm and configure the parameters of the target detection algorithm. In the target detection algorithm, obtain interference features by filtering the image, train the weight matrix, and the target detection algorithm performs target detection through the interference features and the weight matrix to obtain the second target position; S4: Calculate the comprehensive similarity score based on the first target position and the second target position, and use the comprehensive similarity score for screening to obtain potential high-risk targets; S5: Perform position verification on the potential high-risk targets to obtain high-risk targets, determine the wave height, obtain the dynamic safety distance of the high-risk targets based on the wave height, calculate the collision risk index based on the dynamic safety distance and the total processing delay, and issue a collision warning based on the collision risk index.

[0005] According to the collision warning method based on target detection and recognition information provided by the present invention, step S1 further includes: S11: Determine the installation positions of the camera and the receiving device, and install the camera and the receiving device on the target vessel according to the installation positions; S12: The camera takes pictures of the surrounding sea area where there are vessels to obtain the image to be processed, and the receiving device receives the AIS information sent by other vessels and obtains the first target position from the AIS information.

[0006] According to the collision warning method based on target detection and recognition information provided by the present invention, in step S2, determine the number of pixels of the camera and the number of message packets of the AIS information, so as to obtain data processing parameters including the number of pixels, the number of message packets, hardware parameters, and operation complexity; Set up a sea area environment sensor, obtain weather parameters through the sea area environment sensor, and obtain the sea area environment parameters including the main frequency component and the filtering bandwidth control parameters according to the weather parameters.

[0007] According to the collision warning method based on target detection and recognition information provided by the present invention, in step S2, the total processing delay is calculated as follows: where, is the number of pixels, is the number of message packets, is the first operation complexity parameter, is the second operation complexity parameter, is the main frequency of the graphics processing unit, is the main frequency of the programmable array.

[0008] According to the collision warning method based on target detection and recognition information provided by the present invention, in step S2, the filtered image is obtained by the following method: wherein, is to perform Fourier transform on the content in the brackets, is to perform inverse Fourier transform on the content in the brackets, σ is the filtering bandwidth control parameter, u is the spatial frequency in the width direction, v is the spatial frequency in the height direction, is the reference frequency in the width direction, is the reference frequency in the height direction, is the original frequency domain energy.

[0009] According to the collision warning method based on target detection and recognition information provided by the present invention, step S3 further includes: S31: Using the YOLO algorithm as the target detection algorithm, and setting the number of input channels, the size of the convolutional kernel, and the padding coefficient to complete parameter configuration; S32: Extracting features from the filtered image in the target detection algorithm to obtain the interference features, obtaining the weight matrix, and training the weight matrix using the backpropagation algorithm to obtain weighted features through the weight matrix and the interference features; S33: Feeding the weighted features into the detection head of the target detection algorithm, and the target detection algorithm performing target detection on the filtered image to obtain the second target position.

[0010] According to the collision warning method based on target detection and recognition information provided by the present invention, in step S4, the comprehensive similarity score is calculated by the following method: wherein, exp() represents performing exponential function operation on the content in the brackets, is the second position coordinate, is the first position coordinate, represents taking the square root of the sum of the squares of the coordinate differences obtained by subtracting the respective coordinate components of the second position coordinate and the first position coordinate, is the position error tolerance radius, is the second heading angle, is the first heading angle.

[0011] According to the collision warning method based on target detection and recognition information provided by the present invention, step S5 further includes: S51: Performing position verification on the potential high-risk target according to the first target position and the second target position to obtain a high-risk target; S52: Obtain the wave height through the sea area environment sensor, obtain the dynamic safety distance of high-risk targets based on the wave height, calculate the collision risk index based on the dynamic safety distance, determine the delay threshold and the risk index threshold, and issue a collision warning when the total processing delay is less than the delay threshold and the collision risk index is greater than the risk index threshold.

[0012] The present invention also provides a collision warning system based on target detection and recognition information, including: The first target position module: used to deploy a camera and a receiving device on the target ship, the camera captures the image to be processed, and the receiving device receives the AIS information of other ships to obtain the first target position; The filtered image module: used to determine the data processing parameters and the sea area environment parameters, calculate the total processing delay according to the data processing parameters, and perform signal filtering on the image to be processed according to the sea area environment parameters to obtain a filtered image; The second target position module: used to obtain a target detection algorithm and configure parameters for the target detection algorithm, obtain interference features through the filtered image in the target detection algorithm, train the weight matrix, and the target detection algorithm performs target detection through the interference features and the weight matrix to obtain the second target position; The potential high-risk target module: used to calculate the comprehensive similarity score through the first target position and the second target position, and use the comprehensive similarity score for screening to obtain potential high-risk targets; The collision warning module: used to perform position verification on the potential high-risk targets to obtain high-risk targets, determine the wave height, obtain the dynamic safety distance of the high-risk targets based on the wave height, calculate the collision risk index through the dynamic safety distance and the total processing delay, and issue a collision warning according to the collision risk index.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps of the collision warning method based on target detection and recognition information as described in any one of the above.

[0014] One or more of the above technical solutions in the embodiments of the present invention have at least one of the following technical effects: The collision warning method, system and device based on target detection and recognition information provided by the present invention filter the interference caused by sea waves and water mist in the image, effectively eliminating the interference caused by sea waves and water mist, and integrating the information obtained through AIS and the information obtained through image detection to verify the target position, effectively identifying high-risk targets, and performing collision warnings for high-risk targets according to the specific situation of sea waves, so that the method can effectively warn of potential collisions in various situations.

[0015] Additional aspects and advantages of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 It is a schematic flow chart of the collision warning method based on target detection and recognition information provided by the present invention.

[0018] Figure 2 It is a schematic structural diagram of the collision warning system based on target detection and recognition information provided by the present invention.

[0019] Figure 3 It is a schematic structural diagram of the collision warning device based on target detection and recognition information provided by the present invention.

[0020] REFERENCE NUMERALS: 100, First target position module; 200, Filtered image module; 300, Second target position module; 400, Potential high-risk target module; 500, Collision warning module; 810, Processor; 820, Communication interface; 830, Memory; 840, Communication bus. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below. Obviously, the described embodiments are some, but not all, embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. The following embodiments are used to illustrate the present invention, but cannot be used to limit the scope of the present invention.

[0022] In the description of the embodiments of the present invention, it should be noted that the terms "first", "second", and "third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0023] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the embodiments of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0024] The following combines Figures 1 to 3 to describe the specific implementation manners of the present invention Figure 1 is a schematic flowchart of the collision warning method based on target detection and recognition information provided by the present invention. First, a camera captures an image to be processed, and a receiving device receives the AIS information of other vessels to obtain the first target position; then, data processing parameters and sea area environment parameters are determined, the total processing delay is calculated, and the image to be processed is signal-filtered to obtain a filtered image; subsequently, a target detection algorithm is obtained, interference features are obtained, a weight matrix is trained, and the target detection algorithm performs target detection to obtain the second target position; then, the comprehensive similarity score is calculated through the first target position and the second target position for screening to obtain potential high-risk targets; finally, position verification is performed to obtain high-risk targets, the dynamic safety distance of the high-risk targets is obtained, and collision warning is performed according to the collision risk index.

[0025] The present invention provides a collision warning method based on target detection and recognition information, including: S1: Deploy a camera and a receiving device on the target vessel, the camera captures an image to be processed, and the receiving device receives the AIS information of other vessels, thereby obtaining the first target position; Furthermore, the purpose of this stage is to deploy the camera and the receiving device, capture the image to be processed, and the receiving device receives the AIS information, and then obtain the first target position. Specifically, step S1 further includes: S11: Determine the installation positions of the camera and the receiving device, and install the camera and the receiving device on the target vessel according to the installation positions; S12: The camera captures the surrounding sea area where there are vessels to obtain the image to be processed, and the receiving device receives the AIS information sent by other vessels and obtains the first target position from the AIS information.

[0026] For the above steps, the specific implementation in this embodiment is as follows: First, determine the installation positions of the camera and the receiving device. Here, the installation position of the camera needs to be as high as possible and there should be as few other objects around to block it, so as to ensure its shooting range. Moreover, the camera can achieve 360° panoramic coverage and has the characteristics of salt fog resistance and corrosion resistance. For the receiving device used to receive the AIS information sent by other vessels, it is hoped that the electromagnetic interference from other electronic devices can be excluded as much as possible around its installation position, and it is preferably located at positions such as masts to ensure the authenticity and accuracy of the received AIS information and avoid interference as much as possible. In addition, the installation position also needs to provide installation conditions for hardware such as power supply and data transmission interfaces. Subsequently, install the camera and the receiving device at their respective installation positions on the target vessel, that is, the vessel itself.

[0027] Then the camera captures the surrounding sea area where there are vessels, so that an image to be processed including vessels can be obtained. Information such as the positions and headings of the vessels can be obtained from the image to be processed. In addition, the receiving device can also receive the AIS information sent by other vessels. The AIS information contains the positions, headings, models, etc. of other vessels. Therefore, the position information including positions and headings of other vessels, that is, the first target position, can be obtained from the AIS information.

[0028] S2: Determine the data processing parameters and the sea area environment parameters, calculate the total processing delay according to the data processing parameters, and perform signal filtering on the image to be processed according to the sea area environment parameters to obtain a filtered image; Furthermore, the purpose of this stage is to calculate the total processing delay and then perform signal filtering on the image to be processed to obtain a filtered image. Specifically, in step S2, determine the pixel number of the camera and the number of message packets of the AIS information, so as to obtain the data processing parameters including the pixel number, the number of message packets, hardware parameters, and operation complexity; Set up sea area environment sensors, obtain weather parameters through the sea area environment sensors, and obtain the sea area environment parameters including the main frequency component and the filtering bandwidth control parameters according to the weather parameters.

[0029] In step S2, the total processing delay is calculated as follows: where is the pixel number, is the number of message packets, is the first operation complexity parameter, is the second operation complexity parameter, is the main frequency of the graphics processing unit, is the main frequency of the programmable array.

[0030] In step S2, the method for obtaining the filtered image is as follows: Among them, is to perform a Fourier transform on the content in the parentheses, is to perform an inverse Fourier transform on the content in the parentheses, σ is the filter bandwidth control parameter, u is the spatial frequency in the width direction, v is the spatial frequency in the height direction, is the reference frequency in the width direction, is the reference frequency in the height direction, is the original frequency domain energy.

[0031] For the above steps, the specific implementation in this embodiment is as follows: First, it is necessary to obtain the number of pixels of the image to be processed captured by the camera according to the specific parameters of the camera. In addition, it is also necessary to count the number of AIS information packets received by the receiving device at this time and the main frequencies of the graphics processing unit and the programmable array of the computer on the ship as hardware parameters. For the operation complexity, the more complex operations mainly come from the target detection algorithm and calculating the comprehensive similarity score. Therefore, the second operation complexity parameter can be evaluated according to the obtained number of packets and the number of ships included in the image to be processed, and the first operation complexity parameter can be evaluated according to the amount of operations required by the target detection algorithm, so as to obtain the operation complexity.

[0032] Then the total processing delay can be calculated according to the data processing parameters : Among them, is the number of pixels, is the number of packets, is the first operation complexity parameter, is the second operation complexity parameter, is the main frequency of the graphics processing unit, is the main frequency of the programmable array.

[0033] Subsequently, a sea area environment sensor is set up. The sea area environment sensor can collect weather parameters of the sea area where the ship is located, including wave height, wind direction, wave direction, etc. According to the weather parameters, sea area environment parameters including the main frequency component and the filter bandwidth control parameter are obtained. Then the image to be processed can be signal-filtered according to the sea area environment parameters to obtain the filtered image : Among them, is the Fourier transform of the content in the parentheses, is the inverse Fourier transform of the content in the parentheses. σ is a filtering bandwidth control parameter set according to experience, used to control the filtering width. The larger this value, the wider the filtering width. u is the spatial frequency in the width direction of the image to be processed, and v is the spatial frequency in the height direction of the image to be processed. When following the waves, the spatial frequency in the width direction dominates, and vice versa when against the waves. is the reference frequency in the width direction, is the reference frequency in the height direction. The reference frequency in the width direction and the reference frequency in the height direction are determined according to the wave height and the wave direction, and are used to determine the main energy concentration area of the sea wave interference in the frequency domain. is the original frequency domain energy. Here, the way to obtain the original frequency domain energy is to perform a two-dimensional discrete Fourier transform on the original image to be processed, converting the image from the spatial domain to the frequency domain.

[0034] S3: Obtain the target detection algorithm and configure the parameters of the target detection algorithm. In the target detection algorithm, obtain the interference features by filtering the image, train the weight matrix, and the target detection algorithm performs target detection through the interference features and the weight matrix to obtain the second target position; Furthermore, the purpose of this stage is to use the target detection algorithm and train the weight matrix, and then perform target detection to obtain the second target position. Specifically, step S3 further includes: S31: Use the YOLO algorithm as the target detection algorithm, and set the number of input channels, the size of the convolutional kernel, and the padding coefficient to complete the parameter configuration; S32: Extract features from the filtered image in the target detection algorithm to obtain the interference features, obtain the weight matrix, and use the backpropagation algorithm to train the weight matrix to obtain the weighted features through the weight matrix and the interference features; S33: Send the weighted features into the detection head of the target detection algorithm, and the target detection algorithm performs target detection on the filtered image to obtain the second target position.

[0035] For the above steps, the specific implementation method in this embodiment is as follows: First, select the target detection algorithm. Here, the YOLO algorithm is selected as the target detection algorithm. Then, it is also necessary to set the number of input channels, the size of the convolutional kernel, and the padding coefficient. Here, the number of input channels is set to 3, corresponding to the three RGB channels, the number of output channels is set to 16, the size of the convolutional kernel is set to 5×5, and the padding coefficient is set to 2, thus completing the parameter configuration.

[0036] Subsequently, in the target detection algorithm, first, feature extraction is performed on the filtered image in the feature extraction stage to obtain the interference feature X with 16 channels. In addition, a weight matrix M determined according to experience needs to be obtained, and the weight matrix is trained using the backpropagation algorithm to obtain the weighted feature W: wherein, is the sigmoid activation function.

[0037] Subsequently, the weighted feature is fed into the detection head of the target detection algorithm, and the target detection algorithm is used to perform target detection on the filtered image. Then, target detection of the ships in the image to be processed can be performed under the condition of removing the interference caused by the sea waves, and the second target position of the ships, including the position and heading angle of the ships, can be obtained according to the position of the ships in the image after target detection and the position of the own ship. This can improve the target detection accuracy of the targets of white fishing boats in the present invention by 12.6%. It is also necessary to perform pairing according to the second target position to obtain the first target position of the ship. When no matching first target position can be paired according to the second target position, it is considered that the AIS of the ship is turned off or it has forged AIS information, and it is directly marked as a high-risk target.

[0038] S4: Calculate the comprehensive similarity score through the first target position and the second target position, and use the comprehensive similarity score for screening to obtain potential high-risk targets; Furthermore, the purpose of this stage is to calculate the comprehensive similarity score and perform screening to obtain potential high-risk targets. Specifically, in step S4, the comprehensive similarity score is calculated as follows: wherein, exp() represents performing an exponential function operation on the content in the parentheses, is the second position coordinate, is the first position coordinate, represents taking the square root of the sum of the squares of the coordinate differences obtained by subtracting the coordinate components of the second position coordinate and the first position coordinate, is the position error tolerance radius, is the second heading angle, is the first heading angle.

[0039] For the above steps, the specific implementation method in this embodiment is as follows: First, calculate the comprehensive similarity score through the first target position and the second target position : Among them, exp() represents performing an exponential function operation on the content within the parentheses. is the second position coordinate, obtained from the second target position. is the first position coordinate, obtained from the first target position. represents taking the square root of the sum of the squares of the coordinate differences obtained by subtracting the respective coordinate components of the second position coordinate and the first position coordinate. For example, if the first position coordinate is (1, 4) and the second position coordinate is (3, 3), then subtracting 1 from 3 gives -2, subtracting 4 from 3 gives 1. Then, the squares of -2 and 1 are summed, and the square root of the sum is taken. is the position error tolerance radius set according to experience, which is 50m here. is the second course angle obtained from the second target position. is the first course angle obtained from the first target position.

[0040] Subsequently, screening is performed using the comprehensive similarity score, that is, setting a similarity score threshold, which is set to 0.75 here. When the comprehensive similarity score is greater than the similarity score threshold, it can be considered as a potential high-risk target.

[0041] S5: Perform position verification on the potential high-risk target to obtain a high-risk target, determine the wave height, obtain the dynamic safety distance of the high-risk target based on the wave height, calculate the collision risk index through the dynamic safety distance and the total processing delay, and issue a collision warning based on the collision risk index.

[0042] Furthermore, the purpose of this stage is to obtain a high-risk target, determine the wave height, thereby determine the dynamic safety distance, calculate the collision risk index, and issue a collision warning. Specifically, step S5 further includes: S51: Perform position verification on the potential high-risk target according to the first target position and the second target position to obtain a high-risk target; S52: Obtain the wave height through a sea area environment sensor, obtain the dynamic safety distance of the high-risk target based on the wave height, calculate the collision risk index according to the dynamic safety distance, determine the delay threshold and the risk index threshold, and issue a collision warning when the total processing delay is less than the delay threshold and the collision risk index is greater than the risk index threshold.

[0043] For the above steps, the specific implementation methods in this embodiment are as follows: First, perform position verification on the potential high-risk target according to the first target position and the second target position, that is, judge whether the distance between the first target position and the second target position of the potential high-risk target is less than 50m, and whether the difference in course angles is less than 15°. If this condition is met, it is judged as a legal ship; otherwise, it is judged as a non-cooperative high-risk target.

[0044] Subsequently, the wave height and sea state are obtained through the sea area environment sensor, and the dynamic safety distance of the high-risk target is obtained based on the wave height and sea state : where L is the length of the own ship, is the wave height.

[0045] Then, the collision risk index is calculated according to the dynamic safety distance : where TCPA is the time estimated for the own ship to reach the potential collision point after obtaining the potential collision point with the high-risk target using the closest point of approach algorithm, is the speed difference between the own ship and the high-risk target, is the possible maximum speed difference, which is taken as 30 knots here. Determine the delay threshold, and the delay threshold requirements for different collision warning levels are different to avoid false alarms caused by too large a difference between the first target position and the second target position due to the total processing delay. It is also necessary to determine the risk index threshold. Specifically, when the collision risk index is greater than the first risk index threshold 1 and the total processing delay is less than the first delay threshold 50 ms, a first-level collision warning is triggered, that is, an audible and visual warning is given; when the collision risk index is greater than the second risk index threshold 1.5 and the total processing delay is less than the first delay threshold 40 ms, a second-level collision warning is triggered, and an anti-collision suggestion is automatically generated; when the collision risk index is greater than the third risk index threshold 2 and the total processing delay is less than the third delay threshold 30 ms, a third-level collision warning is triggered, and at this time, emergency avoidance measures must be taken.

[0046] When in use, the present invention can effectively reduce the influence of sea waves and give warnings of potential collisions.

[0047] Next, the collision warning device based on target detection and recognition information provided by the present invention is described. The collision warning device based on target detection and recognition information described below can be mutually corresponding and referred to the collision warning method based on target detection and recognition information described above.

[0048] Figure 2 The structural schematic diagram of an example collision warning system based on target detection and recognition information is shown in Figure 2 as shown, and is used to execute the collision warning method based on target detection and recognition information as described above, including: The first target position module 100: used to deploy a camera and a receiving device on the target ship. The camera captures the image to be processed, and the receiving device receives the AIS information of other ships, so as to obtain the first target position; Filtered Image Module 200: It is used to determine data processing parameters and sea area environment parameters, calculate the total processing delay according to the data processing parameters, and perform signal filtering on the image to be processed according to the sea area environment parameters to obtain a filtered image; Second Target Location Module 300: It is used to obtain a target detection algorithm and configure parameters for the target detection algorithm. In the target detection algorithm, interference features are obtained through the filtered image, and a weight matrix is trained. The target detection algorithm performs target detection through the interference features and the weight matrix to obtain the second target location; Potential High-Risk Target Module 400: It is used to calculate a comprehensive similarity score through the first target location and the second target location, and perform screening using the comprehensive similarity score to obtain potential high-risk targets; Collision Warning Module 500: It is used to perform position verification on the potential high-risk targets to obtain high-risk targets, determine the wave height, and obtain the dynamic safety distance of the high-risk targets according to the wave height. Calculate the collision risk index through the dynamic safety distance and the total processing delay, and perform collision warning according to the collision risk index.

[0049] Figure 3 An example of the physical structure diagram of an electronic device is shown in Figure 3 As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830, and a communication bus 840. Among them, the processor 810, the communication interface 820, and the memory 830 complete mutual communication through the communication bus 840. The processor 810 can call the logical instructions in the memory 830 to execute a collision warning method based on target detection and recognition information. The method includes: S1: Deploy a camera and a receiving device on the target ship. The camera captures an image to be processed, and the receiving device receives the AIS information of other ships to obtain the first target location; S2: Determine data processing parameters and sea area environment parameters, calculate the total processing delay according to the data processing parameters, and perform signal filtering on the image to be processed according to the sea area environment parameters to obtain a filtered image; S3: Obtain a target detection algorithm and configure parameters for the target detection algorithm. In the target detection algorithm, interference features are obtained through the filtered image, and a weight matrix is trained. The target detection algorithm performs target detection through the interference features and the weight matrix to obtain the second target location; S4: Calculate a comprehensive similarity score through the first target location and the second target location, and perform screening using the comprehensive similarity score to obtain potential high-risk targets; S5: Perform position verification on the potential high-risk targets to obtain high-risk targets, determine the wave height, obtain the dynamic safety distance of the high-risk targets based on the wave height, calculate the collision risk index through the dynamic safety distance and the total processing delay, and issue a collision warning based on the collision risk index.

[0050] In addition, when the logical instructions in the above-mentioned memory 830 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0051] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative labor.

[0052] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the technical solution, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium such as ROM / RAM, magnetic disks, optical discs, etc., and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments.

[0053] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A collision warning method based on target detection and recognition information, characterized in that Including: S1: Deploy a camera and a receiving device on the target vessel. The camera captures an image to be processed, and the receiving device receives the AIS information of other vessels to obtain a first target position. S2: Determine data processing parameters and sea area environment parameters. Calculate the total processing delay according to the data processing parameters, and perform signal filtering on the image to be processed according to the sea area environment parameters to obtain a filtered image. S3: Obtain a target detection algorithm and configure its parameters. In the target detection algorithm, obtain interference features from the filtered image and train a weight matrix. The target detection algorithm performs target detection through the interference features and the weight matrix to obtain a second target position. S4: Calculate a comprehensive similarity score through the first target position and the second target position, and use the comprehensive similarity score for screening to obtain potential high-risk targets. S5: Perform position verification on the potential high-risk targets to obtain high-risk targets, determine the wave height, and obtain the dynamic safety distance of the high-risk targets according to the wave height. Calculate the collision risk index through the dynamic safety distance and the total processing delay, and issue a collision warning according to the collision risk index.

2. The collision warning method based on target detection and recognition information according to claim 1, wherein Step S1 further includes: S11: Determine the installation positions of the camera and the receiving device, and install the camera and the receiving device on the target vessel according to the installation positions. S12: The camera captures the surrounding sea area where there are vessels to obtain the image to be processed, and the receiving device receives the AIS information sent by other vessels and obtains the first target position from the AIS information.

3. The collision warning method based on target detection and recognition information according to claim 1, characterized in that In step S2, determine the number of pixels of the camera and the number of AIS information messages to obtain data processing parameters including the number of pixels, the number of messages, hardware parameters, and operation complexity. Set up a sea area environment sensor, obtain weather parameters through the sea area environment sensor, and obtain the sea area environment parameters including the main frequency component and the filter bandwidth control parameter according to the weather parameters.

4. The collision warning method based on target detection and recognition information according to claim 3, characterized in that In step S2, the total processing delay is calculated as follows: wherein, is the number of pixels, is the number of messages, is the first operation complexity parameter, is the second operation complexity parameter, is the main frequency of the graphics processing unit, is the main frequency of the programmable array.

5. The collision warning method based on target detection and recognition information according to claim 3, wherein In step S2, the filtered image is obtained by the following method: Among them, is to perform Fourier transform on the content in the parentheses, is to perform inverse Fourier transform on the content in the parentheses, σ is the filtering bandwidth control parameter, u is the spatial frequency in the width direction, and v is the spatial frequency in the height direction, is the reference frequency in the width direction, is the reference frequency in the height direction, is the original frequency domain energy.

6. The collision warning method based on target detection and recognition information according to claim 1, wherein Step S3 further includes: S31: Use the YOLO algorithm as the target detection algorithm, and set the number of input channels, the convolutional kernel size, and the padding coefficient to complete parameter configuration. S32: Extract features from the filtered image in the target detection algorithm to obtain the interference features, obtain the weight matrix, and use the backpropagation algorithm to train the weight matrix to obtain weighted features through the weight matrix and the interference features. S33: Send the weighted features into the detection head of the target detection algorithm, and the target detection algorithm performs target detection on the filtered image to obtain the second target position.

7. The collision warning method based on object detection and recognition information according to claim 1, wherein In step S4, the comprehensive similarity score is calculated as follows: where exp() represents the exponential function operation on the content within the parentheses, is the second position coordinate, is the first position coordinate, represents taking the square root of the sum of the squares of the coordinate differences obtained by subtracting the respective coordinate components of the second position coordinate and the first position coordinate, is the position error tolerance radius, is the second heading angle, is the first heading angle.

8. The collision warning method based on target detection and recognition information according to claim 1, characterized in that Step S5 further includes: S51: Perform position verification on the potential high-risk targets according to the first target position and the second target position to obtain high-risk targets. S52: Obtain the wave height through the sea area environment sensor, obtain the dynamic safety distance of high-risk targets based on the wave height, calculate the collision risk index according to the dynamic safety distance, determine the delay threshold and the risk index threshold, and issue a collision warning when the total processing delay is less than the delay threshold and the collision risk index is greater than the risk index threshold.

9. A collision warning system based on target detection and recognition information, for performing the collision warning method based on target detection and recognition information according to any one of claims 1 to 8, characterized in that, Including: The first target position module: used to deploy a camera and a receiving device on the target ship, the camera captures the image to be processed, and the receiving device receives the AIS information of other ships, so as to obtain the first target position; The filtered image module: used to determine the data processing parameters and the sea area environment parameters, calculate the total processing delay according to the data processing parameters, and perform signal filtering on the image to be processed according to the sea area environment parameters to obtain a filtered image; The second target position module: used to obtain the target detection algorithm and configure the parameters of the target detection algorithm, obtain interference features through the filtered image in the target detection algorithm, train the weight matrix, and the target detection algorithm performs target detection through the interference features and the weight matrix to obtain the second target position; The potential high-risk target module: used to calculate the comprehensive similarity score through the first target position and the second target position, and use the comprehensive similarity score for screening to obtain potential high-risk targets; The collision warning module: used to perform position verification on the potential high-risk targets to obtain high-risk targets, determine the wave height, and obtain the dynamic safety distance of the high-risk targets according to the wave height, calculate the collision risk index through the dynamic safety distance and the total processing delay, and issue a collision warning according to the collision risk index.

10. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the collision warning method based on target detection and recognition information according to any one of claims 1 to 8.

Images