A tracking and connection method and device for marine targets
By marking and processing the trajectory data of maritime targets, and using the maritime target tracking model to judge the continuation probability, the problem of poor continuation effect of maritime targets is solved, the accuracy and timeliness of continuation are improved, and it is suitable for modern marine construction.
Patent Information
- Application Number
- CN202311732933.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-12-15
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2043-12-15
AI Technical Summary
The existing maritime target connection technology has poor effect and cannot effectively deal with the problems of unstable maritime signal and excessive stray targets, resulting in interruption of targets and affecting modern ocean construction.
By receiving the trajectory data of the maritime target, marking, grouping, processing and inputting a preset maritime target tracking model, the probability value is calculated to determine whether the trajectory data can be continued, and then conducting continuation processing.
It improves the accuracy of maritime target connection, reduces the data processing volume and the number of stray targets, and meets the timeliness requirements of modern marine radar networks.
Smart Images

Figure CN117893565B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of target tracking, and also to a method and device for tracking and connecting a target at sea. Background Art
[0002] In modern ocean construction, the amount of marine target data is large, and there are many stray targets. Due to unstable marine signals, occlusion, and abnormal reasons such as boundaries that cannot be covered by some algorithms, marine target interruptions may occur. Too many stray targets and target interruptions will cause many big data algorithms to be inapplicable, and will also lead to too many targets, which will seriously affect modern ocean construction. Research on marine target continuation can effectively reduce the system's data processing volume, reduce the number of marine targets, and reduce the amount of data on stray targets. Therefore, research on marine target continuation is an indispensable step in modern ocean construction.
[0003] The existing maritime target connection technology is mainly aimed at ships with the automatic ship identification system turned on, and the connection is based on the characteristics such as the water mobile service identification code, ship name and distance. However, in fact, many targets do not actually turn on the automatic ship identification system, which will result in a large number of targets not being connected. In addition, the existing technology often adopts an offline method to connect to the target, and the data volume is limited. In the construction of modern marine radar networks, the data volume is large and the timeliness requirements are high. It is difficult for existing technologies to meet the requirements. Summary of the invention
[0004] The technical problem to be solved by the present invention is to provide a method and device for tracking and connecting a target at sea, so as to solve the problem of poor target connection effect in the prior art.
[0005] In order to solve the above technical problems, the technical solution of the present invention is as follows:
[0006] A first aspect of the present invention provides a method for tracking and connecting a target at sea, comprising:
[0007] Receive trajectory data of maritime targets;
[0008] Marking the trajectory data according to preset conditions to obtain marked trajectory data;
[0009] Grouping the marked trajectory data to obtain a grouping result;
[0010] Processing the labeled trajectory data according to the grouping result to obtain feature data;
[0011] Inputting the characteristic data into a preset marine target tracking model to obtain a probability value;
[0012] The trajectory data is processed in a continuous manner according to the probability value, the preset continuous value and the grouping result.
[0013] Optionally, the trajectory data is marked according to a preset condition to obtain the marked trajectory data, including:
[0014] When the duration of the trajectory data is zero, marking the trajectory data to obtain newly generated flow data;
[0015] According to the trigger time of the timer, the trajectory data is marked to obtain the disappearing flow data; wherein the trigger time of the timer is calculated by the formula:
[0016] z = (currentTime + outTime) / combineTime, and t = z × combineTime;
[0017] Among them, z is the integer part, currentTime is the current time, outTime is the timeout time, combineTime is the merge time, and t is the trigger time of the timer.
[0018] Optionally, grouping the marked trajectory data to obtain a grouping result includes:
[0019] Performing code value calculation on the marked trajectory data to obtain a code value;
[0020] The marked trajectory data are grouped according to the encoding value to obtain a grouping result.
[0021] Optionally, the marked trajectory data is processed according to the grouping result to obtain feature data, including:
[0022] According to the newly generated flow data, the disappeared flow data and the formula The relative distance error is calculated; where e is the relative distance error, v is the speed of the disappearing target at the time of disappearance, t0 is the time difference between the starting point of the newly generated target and the time when the disappearing target disappears, and d is the distance between the starting point of the newly generated target and the time when the disappearing target disappears;
[0023] If the relative distance error is less than a preset distance value, the marked trajectory data is processed according to the grouping result to obtain feature data.
[0024] Optionally, the marked trajectory data is processed according to the grouping result to obtain feature data, including:
[0025] Based on the labeled trajectory data and formula of the same group:
[0026] v1=(lon2-lon1,lat2-lat1)
[0027] v2=(lon3-lon2,lat3-lat2)
[0028] Get feature data;
[0029] Among them, lon1 is the longitude of the disappearing target at the moment before it disappears, lat1 is the latitude of the disappearing target at the moment before it disappears, lon2 is the longitude of the disappearing target at the moment it disappears, lat2 is the latitude of the disappearing target at the moment it disappears, lon3 is the longitude of the starting point of the newly generated target, lat3 is the latitude of the starting point of the newly generated target, v1 is the vector between the disappearance moment of the disappearing target and the moment before it disappears, v2 is the vector between the starting point of the newly generated target and the disappearance moment of the disappearing target, and theta is the angle between the direction of the disappearing target and the direction of the newly generated target.
[0030] Optionally, inputting the characteristic data into a preset marine target tracking model to obtain a probability value includes:
[0031] Sorting the characteristic data to obtain ordered characteristic data;
[0032] Performing histogram division on the ordered feature data to obtain a plurality of histograms;
[0033] Constructing a decision tree for the plurality of histograms to obtain a decision tree;
[0034] According to the decision tree, a probability value is obtained.
[0035] Optionally, the trajectory data is subsequently processed, including:
[0036] If the probability value is greater than a preset connection value, the trajectory data of the same group are merged.
[0037] A second aspect of the present invention provides a tracking and connection device for a marine target, comprising:
[0038] A receiving module, used for receiving trajectory data of maritime targets;
[0039] A marking module, used for marking the trajectory data according to preset conditions to obtain marked trajectory data;
[0040] A grouping module, used for grouping the marked trajectory data to obtain a grouping result;
[0041] A processing module, used for processing the marked trajectory data according to the grouping result to obtain feature data;
[0042] An input module, used for inputting the characteristic data into a preset marine target tracking model to obtain a probability value;
[0043] A connection module is used to perform connection processing on the trajectory data according to the probability value, the preset connection value and the grouping result.
[0044] According to a third aspect of the present invention, a computing device is provided, comprising: a processor and a memory storing a computer program, wherein when the computer program is executed by the processor, the method according to the first aspect is executed.
[0045] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions, which, when executed on a computer, causes the computer to execute the method described in the first aspect.
[0046] The above solution of the present invention includes at least the following beneficial effects:
[0047] The above-mentioned scheme of the present invention can effectively identify whether the maritime target disappears and whether the maritime target is newly generated by marking the trajectory data of the maritime target, and then use the preset maritime target tracking model to determine the probability value that the trajectory data can be continued, and then continue to process the trajectory data, which greatly improves the accuracy of maritime target continuation. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] Figure 1 It is a flow chart of a method for tracking and connecting a marine target in an embodiment of the present invention;
[0049] Figure 2 It is a flowchart of a specific embodiment of a method for tracking and connecting a marine target in an embodiment of the present invention;
[0050] Figure 3 It is a schematic structural diagram of a tracking and connecting device for a marine target in an embodiment of the present invention. DETAILED DESCRIPTION
[0051] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present invention and to enable the scope of the present invention to be fully communicated to those skilled in the art.
[0052] like Figure 1 As shown, an embodiment of the present invention provides a tracking and connection method for a marine target, comprising:
[0053] Step 11, receiving trajectory data of the sea target;
[0054] Step 12, marking the trajectory data according to preset conditions to obtain marked trajectory data;
[0055] Step 13, grouping the marked trajectory data to obtain a grouping result;
[0056] Step 14, processing the marked trajectory data according to the grouping result to obtain feature data;
[0057] Step 15, inputting the characteristic data into a preset marine target tracking model to obtain a probability value;
[0058] Step 16: performing connection processing on the trajectory data according to the probability value, the preset connection value and the grouping result.
[0059] The method for tracking and continuing a maritime target proposed in an embodiment of the present invention can effectively identify whether a maritime target has disappeared and whether a maritime target is newly generated by marking the trajectory data of the maritime target, and then use a preset maritime target tracking model to determine the probability value that the trajectory data can be continued, and then continue processing the trajectory data, thereby greatly improving the accuracy of maritime target connection.
[0060] In this embodiment of the present invention, the marine target may be a ship, or other machines or organisms active at sea; the trajectory data of the marine target generally includes the following contents: heading, speed, longitude and latitude, duration, etc. The trajectory data of the marine target may be collected through AIS targets, Beidou targets, radar targets, photoelectric targets, etc.
[0061] In an optional embodiment of the present invention, step 12 includes:
[0062] When the duration of the trajectory data is zero, marking the trajectory data to obtain newly generated flow data;
[0063] According to the trigger time of the timer, the trajectory data is marked to obtain the disappearing flow data; wherein the trigger time of the timer is calculated by the formula:
[0064] z = (currentTime + outTime) / combineTime, and t = z × combineTime;
[0065] Among them, z is the integer part, currentTime is the current time, outTime is the timeout time, combineTime is the merge time, and t is the trigger time of the timer.
[0066] If the duration of the trajectory data is zero, it means that it is a newly generated maritime target. The trajectory data is the first data of this target, and this trajectory data is marked as newly generated stream data; when the timer is triggered, it means that the maritime target has not generated new trajectory data for a long time, and this target may have disappeared. At this time, if the duration of the trajectory data exceeds the preset value, such as the preset value of 2 minutes, the trajectory data is marked as disappeared stream data; marking the trajectory data as newly generated stream data / disappeared stream data is convenient for judging whether the maritime target is a maritime target that has been tracked, or a newly generated target after the maritime target that has been tracked is lost, and then the trajectory data is processed continuously.
[0067] In an optional embodiment of the present invention, step 13 includes:
[0068] Step 131, calculating the code value of the marked trajectory data to obtain the code value;
[0069] Step 132: group the marked trajectory data according to the coding value to obtain a grouping result.
[0070] The address coding value of the marked trajectory data is calculated. Specifically, the marked trajectory data can be encoded into a character string. For example, the address coding method can be used to calculate the longitude and latitude string values of the disappearing flow data and the longitude and latitude string values of the newly generated flow data respectively; the newly generated target not only calculates the longitude and latitude string codes of the current data, but also calculates the string codes of the 8 points in its field. The steps for calculating the address coding value of the marked trajectory data are as follows: the latitude range (-90, 90) is divided into the first latitude interval (-90, 0) and the second latitude interval (0, 90). If the latitude of the marked trajectory data is in the first interval, the code is 0, otherwise The code is 1, and the latitude interval where the latitude of the marked trajectory data falls is evenly divided into two latitude intervals for encoding, and so on, until the accuracy meets the requirements; similarly, the longitude is processed similarly, and the longitude range (-180, 180) is subdivided in turn for encoding; the longitude code and the latitude code are combined, and the odd-numbered bits are placed in the latitude, and the even-numbered bits are placed in the longitude to form a new string of binary codes, and encoded according to Base32 (a scheme that uses 32 printable characters AZ and numbers 2-7 to encode arbitrary byte data) to obtain the longitude and latitude code values; the newly generated flow data and the disappearing flow data are grouped according to the longitude and latitude code values. The purpose of calculating the address code value and grouping the marked trajectory data is to make the trajectory data with similar distances in the same group as much as possible, so as to facilitate the judgment of whether the trajectory data in the same group are generated by the same marine target.
[0071] In an optional embodiment of the present invention, step 14 includes:
[0072] Step 141, based on the newly generated stream data, the disappeared stream data and the formula The relative distance error is calculated; where e is the relative distance error, v is the speed of the disappearing target at the time of disappearance, t0 is the time difference between the starting point of the newly generated target and the time when the disappearing target disappears, and d is the distance between the starting point of the newly generated target and the time when the disappearing target disappears;
[0073] Step 142: If the relative distance error is less than a preset distance value, the marked trajectory data is processed according to the grouping result to obtain feature data.
[0074] Specifically, first calculate the time difference t0 between the starting point of the newly generated target and the disappearance time of the disappeared target. If t0 is less than or equal to 30 minutes, calculate the distance d between the starting point of the newly generated target and the disappearance time of the disappeared target. If d is less than 500 meters, calculate the relative distance error e. The preset distance value can be 0.3. If the relative distance error is less than 0.3, process the marked trajectory data. Through the above steps, it can be determined whether the trajectory data of the same group meets the time-space position relationship, thereby improving the efficiency of the trajectory data of the maritime target.
[0075] In an optional embodiment of the present invention, step 14 includes:
[0076] Step 143, based on the labeled trajectory data and formula of the same group:
[0077] v1=(lon2-lon1,lat2-lat1)
[0078] v2=(lon3-lon2,lat3-lat2)
[0079] Get feature data;
[0080] Among them, lon1 is the longitude of the disappearing target at the moment before it disappears, lat1 is the latitude of the disappearing target at the moment before it disappears, lon2 is the longitude of the disappearing target at the moment it disappears, lat2 is the latitude of the disappearing target at the moment it disappears, lon3 is the longitude of the starting point of the newly generated target, lat3 is the latitude of the starting point of the newly generated target, v1 is the vector between the disappearance moment of the disappearing target and the moment before it disappears, v2 is the vector between the starting point of the newly generated target and the disappearance moment of the disappearing target, and theta is the angle between the direction of the disappearing target and the direction of the newly generated target.
[0081] Specifically, the characteristic data include time difference, distance, heading difference, angle, speed difference, etc. The time difference, distance, heading difference, and speed difference can be obtained by subtracting the newly generated flow data from the disappearing flow data. For example, the distance is calculated by the newly generated flow data, the disappearing flow data, and the Euclidean distance formula, and the speed difference is the speed in the newly generated flow data minus the speed in the disappearing flow data.
[0082] In an optional embodiment of the present invention, step 15 includes:
[0083] Step 151, sorting the feature data to obtain ordered feature data;
[0084] Step 152, performing histogram division on the ordered feature data to obtain a plurality of histograms;
[0085] Step 153, constructing a decision tree for the plurality of histograms to obtain a decision tree;
[0086] Step 154, obtaining a probability value according to the decision tree.
[0087] In an optional embodiment of the present invention, step 16, performing subsequent processing on the trajectory data, includes:
[0088] If the probability value is greater than a preset connection value, the trajectory data of the same group are merged.
[0089] Specifically, the probability value is the probability that the newly generated flow data and the disappeared flow data are the trajectory data of the same maritime target. If the probability value is greater than the preset continuation value, it means that the newly generated flow data and the disappeared flow data are the trajectory data of the same maritime target. At this time, the newly generated flow data is merged into the maritime target described in the disappeared flow data for storage.
[0090] like Figure 2 As shown, a specific embodiment of the method for tracking and connecting a target at sea according to an embodiment of the present invention is:
[0091] Obtain historical trajectory data from the data warehouse, select 100,000 pieces of accurate data as positive samples, and 100,000 pieces of incorrect data as negative samples. The historical trajectory data mainly contains heading, speed, longitude and latitude information. Clean the trajectory data and filter the sample data with obviously unreasonable heading, speed, longitude and latitude. Calculate the time difference, distance, heading difference, angle, speed difference, and length difference between the two targets before and after the connection (disappearing target and newly generated target) as feature data. Standardize the feature data (such as normalization) to accelerate model convergence. Split the processed sample feature data, 80% as the training set and 20% as the test set. Use the training set to train multiple preset network models. During the training process, adjust the parameters to optimize the effect of each model. Select the optimal model as the preset maritime target tracking model based on the effect of the model in the test set. Save the model file to the distributed file system and update the model file incrementally.
[0092] Receive the trajectory data of the maritime target. The trajectory data mainly includes heading, speed, longitude and latitude, and duration information. Preprocess and filter the data to remove the trajectory data containing virtual data tags and static target data tags. Thin the trajectory data in the time dimension. The radar target updates a point every 3 seconds. Too many targets will cause service backlogs, so further data compression is required and one point is extracted every minute. Each maritime target will have a unique ID (such as a digital code). The trajectory data is grouped according to the target ID to ensure that the trajectory data with the same ID is in the same group. When a new data comes in for the target ID, determine whether the previous timer is empty. If not, delete the previously registered timer and register a new timer. Here, the steps to register the timer are: calculate the sum of the current time and the timeout time and divide the combined time, and take the integer part. The formula is as follows:
[0093] z=(currentTime+outTime) / combineTime;
[0094] Calculate the timer trigger time. The timer trigger time is equal to the integer part of step 1 multiplied by the merge time. The formula is as follows:
[0095] t=z*combineTime
[0096] Among them, z is the integer part, currentTime is the current time, outTime is the timeout time, combineTime is the merge time, and t is the trigger time of the timer.
[0097] Determine whether the duration of the new trajectory data is 0. If so, it means that the target is newly generated and this data is the first data of this target. Mark this data as the newly generated flow data. When the timer is triggered, it means that this target has not received new data for a long time and this target may have disappeared. When the timer is triggered, perform the following steps: Determine whether the duration of the target is greater than 2 minutes (2 minutes is the preset value and can be set as needed); Mark the trajectory data with a duration greater than 2 minutes as disappeared flow data.
[0098] Get the newly generated stream and the disappearing stream, and calculate the longitude and latitude string value of the disappearing point of the disappearing stream and the longitude and latitude string value of the starting point of the newly generated stream based on the address code. The newly generated target not only calculates the longitude and latitude string code of the current data, but also calculates the string code of the 8 points in its field. The calculation steps of the string code value are as follows:
[0099] The latitude range (-90, 90) is divided into the first latitude interval (-90, 0) and the second latitude interval (0, 90). If the latitude of the trajectory point is in the first interval, it is encoded as 0, otherwise it is encoded as 1. The latitude interval where the latitude of the trajectory point falls is evenly divided into two latitude intervals for encoding, and so on, until the accuracy meets the requirements; similarly, the longitude is processed similarly, and the longitude range (-180, 180) is subdivided and encoded in turn; the longitude code and the latitude code are combined, the odd bits are placed in the latitude, and the even bits are placed in the longitude, to form a new string of binary codes, and encoded according to Base32 to obtain the longitude and latitude code values, that is, the string code values; the newly generated flow data and the disappeared flow data are grouped according to the longitude and latitude string code values.
[0100] The disappearing flow data and the newly generated flow data of the same group are connected at intervals. The following steps are performed during the interval connection: Calculate the time difference t0 between the starting point of the newly generated target and the end point of the disappeared target. If t0 is less than or equal to 30 minutes, proceed to the next step, otherwise end; Use the Euclidean distance formula to calculate the distance d between the starting point of the newly generated target and the end point of the disappeared target. If d is less than 500 meters, proceed to the next step, otherwise calculate the relative distance error. The formula for the relative distance error is as follows: If the relative error is less than 0.3, proceed to the next step, otherwise end, where v is the speed of the disappearing target terminal, e is the distance relative error, v is the speed of the disappearing target at the time of disappearance, t0 is the time difference between the starting point of the newly generated target and the disappearance of the disappearing target, and d is the distance between the starting point of the newly generated target and the disappearance of the disappearing target; calculate the angle between the direction of the disappearing target and the direction of the newly generated target, and the angle calculation formula is as follows:
[0101] Calculate the vector of the disappearing target at the moment it disappears and the moment before it disappears:
[0102] v1=(lon2-lon1,lat2-lat1)
[0103] Calculate the vector between the starting point of the newly generated target and the disappearance time of the disappeared target:
[0104] v2=(lon3-lon2,lat3-lat2)
[0105] Calculate the angle between the direction of the disappeared target and the direction of the newly generated target:
[0106] theta=(v1*v2) / |v1||v2|
[0107] Among them, lon1 is the longitude of the disappearing target at the moment before it disappears, lat1 is the latitude of the disappearing target at the moment before it disappears, lon2 is the longitude of the disappearing target at the moment it disappears, lat2 is the latitude of the disappearing target at the moment it disappears, lon3 is the longitude of the starting point of the newly generated target, lat3 is the latitude of the starting point of the newly generated target, v1 is the vector between the disappearance moment of the disappearing target and the moment before it disappears, v2 is the vector between the starting point of the newly generated target and the disappearance moment of the disappearing target, and theta is the angle between the direction of the disappearing target and the direction of the newly generated target.
[0108] After obtaining the characteristic data composed of time difference, distance, heading difference, angle, speed difference, etc., the preset maritime target tracking model is called for prediction, and the probability that the newly generated flow data and the disappearing flow data in the same group are the trajectory data of the same maritime target is calculated. If the probability value is greater than the preset continuation value, it means that the newly generated flow data and the disappearing flow data are the trajectory data of the same maritime target. At this time, the newly generated flow data is incorporated into the maritime target described by the disappearing flow data for storage.
[0109] The method for tracking and connecting a maritime target in an embodiment of the present invention connects the maritime target based on the trajectory data of the maritime target and a preset maritime target tracking model. The method can process massive data, and the timeliness is also guaranteed. It can process multi-source data, avoid the insufficiency of single data, and improve the accuracy of connection. By marking the trajectory data, it is also possible to effectively identify whether the target disappears and whether the target is newly generated, thereby improving the accuracy of target connection.
[0110] like Figure 3 As shown, an embodiment of the present invention provides a tracking and connecting device 100 for a marine target, comprising:
[0111] A receiving module 101 is used to receive trajectory data of a marine target;
[0112] A marking module 102, used to mark the trajectory data according to preset conditions to obtain marked trajectory data;
[0113] A grouping module 103, used to group the marked trajectory data to obtain a grouping result;
[0114] A processing module 104 is used to process the marked trajectory data according to the grouping result to obtain feature data;
[0115] An input module 105 is used to input the characteristic data into a preset marine target tracking model to obtain a probability value;
[0116] The connection module 106 is used to perform connection processing on the trajectory data according to the probability value, the preset connection value and the grouping result.
[0117] Optionally, the trajectory data is marked according to a preset condition to obtain the marked trajectory data, including:
[0118] When the duration of the trajectory data is zero, marking the trajectory data to obtain newly generated flow data;
[0119] According to the trigger time of the timer, the trajectory data is marked to obtain the disappearing flow data; wherein the trigger time of the timer is calculated by the formula:
[0120] z = (currentTime + outTime) / combineTime, and t = z × combineTime;
[0121] Among them, z is the integer part, currentTime is the current time, outTime is the timeout time, combineTime is the merge time, and t is the trigger time of the timer.
[0122] Optionally, grouping the marked trajectory data to obtain a grouping result includes:
[0123] Performing code value calculation on the marked trajectory data to obtain a code value;
[0124] The marked trajectory data are grouped according to the encoding value to obtain a grouping result.
[0125] Optionally, the marked trajectory data is processed according to the grouping result to obtain feature data, including:
[0126] According to the newly generated flow data, the disappeared flow data and the formula The relative distance error is calculated; where e is the relative distance error, v is the speed of the disappearing target at the time of disappearance, t0 is the time difference between the starting point of the newly generated target and the time when the disappearing target disappears, and d is the distance between the starting point of the newly generated target and the time when the disappearing target disappears;
[0127] If the relative distance error is less than a preset distance value, the marked trajectory data is processed according to the grouping result to obtain feature data.
[0128] Optionally, the marked trajectory data is processed according to the grouping result to obtain feature data, including:
[0129] Based on the labeled trajectory data and formula of the same group:
[0130] v1=(lon2-lon1,lat2-lat1)
[0131] v2=(lon3-lon2,lat3-lat2)
[0132] Get feature data;
[0133] Among them, lon1 is the longitude of the disappearing target at the moment before it disappears, lat1 is the latitude of the disappearing target at the moment before it disappears, lon2 is the longitude of the disappearing target at the moment it disappears, lat2 is the latitude of the disappearing target at the moment it disappears, lon3 is the longitude of the starting point of the newly generated target, lat3 is the latitude of the starting point of the newly generated target, v1 is the vector between the disappearance moment of the disappearing target and the moment before it disappears, v2 is the vector between the starting point of the newly generated target and the disappearance moment of the disappearing target, and theta is the angle between the direction of the disappearing target and the direction of the newly generated target.
[0134] Optionally, inputting the characteristic data into a preset marine target tracking model to obtain a probability value includes:
[0135] Sorting the characteristic data to obtain ordered characteristic data;
[0136] Performing histogram division on the ordered feature data to obtain a plurality of histograms;
[0137] Constructing a decision tree for the plurality of histograms to obtain a decision tree;
[0138] According to the decision tree, a probability value is obtained.
[0139] Optionally, the trajectory data is subsequently processed, including:
[0140] If the probability value is greater than a preset connection value, the trajectory data of the same group are merged.
[0141] The tracking and connection device for maritime targets proposed in the embodiment of the present invention can effectively identify whether the maritime target has disappeared and whether the maritime target is newly generated by marking the trajectory data of the maritime target, and then use the preset maritime target tracking model to determine the probability value that the trajectory data can be connected, and then continue the trajectory data, thereby greatly improving the accuracy of maritime target connection.
[0142] It should be noted that the device is a device corresponding to the above method, and all implementations in the above method embodiment are applicable to the embodiment of the device and can achieve the same technical effect. This embodiment will not be repeated.
[0143] The embodiment of the present invention further provides a computing device, comprising: a processor, and a memory storing a computer program, wherein when the computer program is executed by the processor, the method described in any one of the above embodiments is executed. All implementations in the above method embodiments are applicable to the embodiments of the device, and can achieve the same technical effects. They will not be described in detail in this embodiment.
[0144] The embodiment of the present invention further provides a computer-readable storage medium on which instructions are stored. When the instructions are executed on a computer, the computer executes the method as described in any one of the above embodiments. All implementations in the above method embodiments are applicable to the embodiments of the device and can achieve the same technical effects. They will not be described in detail in this embodiment.
[0145] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A method for tracking and connecting a target at sea, characterized in that: include: Receive trajectory data of maritime targets; Marking the trajectory data according to preset conditions to obtain marked trajectory data; Grouping the marked trajectory data to obtain a grouping result; Processing the labeled trajectory data according to the grouping result to obtain feature data; Inputting the characteristic data into a preset marine target tracking model to obtain a probability value; Performing connection processing on the trajectory data according to the probability value, the preset connection value and the grouping result; The step of marking the trajectory data according to a preset condition to obtain the marked trajectory data includes: When the duration of the trajectory data is zero, marking the trajectory data to obtain newly generated flow data; According to the trigger time of the timer, the trajectory data is marked to obtain the disappearing flow data; wherein the trigger time of the timer is calculated by the formula: z = (currentTime + outTime) / combineTime, and t = z × combineTime; Where z is the integer part, currentTime is the current time, outTime is the timeout time, combineTime is the merge time, and t is the trigger time of the timer; The marked trajectory data is processed according to the grouping result to obtain feature data, including: According to the newly generated flow data, the disappeared flow data and the formula The relative distance error is calculated; where e is the relative distance error, v is the speed of the disappearing target at the time of disappearance, t0 is the time difference between the starting point of the newly generated target and the time when the disappearing target disappears, and d is the distance between the starting point of the newly generated target and the time when the disappearing target disappears; If the relative distance error is less than a preset distance value, processing the marked trajectory data according to the grouping result to obtain feature data; The marked trajectory data is processed according to the grouping result to obtain feature data, including: Based on the labeled trajectory data and formula of the same group: v1=(lon2-lon1,lat2-lat1) v2=(lon3-lon2,lat3-lat2) Get feature data; Among them, lon1 is the longitude of the disappearing target at the moment before it disappears, lat1 is the latitude of the disappearing target at the moment before it disappears, lon2 is the longitude of the disappearing target at the moment it disappears, lat2 is the latitude of the disappearing target at the moment it disappears, lon3 is the longitude of the starting point of the newly generated target, lat3 is the latitude of the starting point of the newly generated target, v1 is the vector between the disappearance moment of the disappearing target and the moment before it disappears, v2 is the vector between the starting point of the newly generated target and the disappearance moment of the disappearing target, and theta is the angle between the direction of the disappearing target and the direction of the newly generated target; The process of processing the trajectory data includes: If the probability value is greater than a preset connection value, the trajectory data of the same group are merged; The probability value is the probability that the newly generated flow data and the disappeared flow data are the trajectory data of the same marine target.
2. The method for tracking and connecting a target at sea according to claim 1, characterized in that: The marked trajectory data are grouped to obtain a grouping result, including: Performing code value calculation on the marked trajectory data to obtain a code value; The marked trajectory data are grouped according to the encoding value to obtain a grouping result.
3. The method for tracking and connecting a target at sea according to claim 1, characterized in that: The characteristic data is input into a preset marine target tracking model to obtain a probability value, including: Sorting the characteristic data to obtain ordered characteristic data; Performing histogram division on the ordered feature data to obtain multiple histograms; Constructing a decision tree for the plurality of histograms to obtain a decision tree; According to the decision tree, a probability value is obtained.
4. A tracking and connection device for a marine target, characterized in that: include: A receiving module, used for receiving trajectory data of maritime targets; A marking module, used for marking the trajectory data according to preset conditions to obtain marked trajectory data; A grouping module, used for grouping the marked trajectory data to obtain a grouping result; A processing module, used for processing the marked trajectory data according to the grouping result to obtain feature data; An input module, used for inputting the characteristic data into a preset marine target tracking model to obtain a probability value; A connection module, used for performing connection processing on the trajectory data according to the probability value, the preset connection value and the grouping result; The step of marking the trajectory data according to a preset condition to obtain the marked trajectory data includes: When the duration of the trajectory data is zero, marking the trajectory data to obtain newly generated flow data; According to the trigger time of the timer, the trajectory data is marked to obtain the disappearing flow data; wherein the trigger time of the timer is calculated by the formula: z = (currentTime + outTime) / combineTime, and t = z × combineTime; Where z is the integer part, currentTime is the current time, outTime is the timeout time, combineTime is the merge time, and t is the trigger time of the timer; The marked trajectory data is processed according to the grouping result to obtain feature data, including: According to the newly generated flow data, the disappeared flow data and the formula The relative distance error is calculated; where e is the relative distance error, v is the speed of the disappearing target at the time of disappearance, t0 is the time difference between the starting point of the newly generated target and the time when the disappearing target disappears, and d is the distance between the starting point of the newly generated target and the time when the disappearing target disappears; If the relative distance error is less than a preset distance value, processing the marked trajectory data according to the grouping result to obtain feature data; The marked trajectory data is processed according to the grouping result to obtain feature data, including: Based on the labeled trajectory data and formula of the same group: v1=(lon2-lon1,lat2-lat1) v2=(lon3-lon2,lat3-lat2) Get feature data; Among them, lon1 is the longitude of the disappearing target at the moment before it disappears, lat1 is the latitude of the disappearing target at the moment before it disappears, lon2 is the longitude of the disappearing target at the moment it disappears, lat2 is the latitude of the disappearing target at the moment it disappears, lon3 is the longitude of the starting point of the newly generated target, lat3 is the latitude of the starting point of the newly generated target, v1 is the vector between the disappearance moment of the disappearing target and the moment before it disappears, v2 is the vector between the starting point of the newly generated target and the disappearance moment of the disappearing target, and theta is the angle between the direction of the disappearing target and the direction of the newly generated target; The process of processing the trajectory data includes: If the probability value is greater than a preset connection value, the trajectory data of the same group are merged; The probability value is the probability that the newly generated flow data and the disappeared flow data are the trajectory data of the same marine target.
5. A computing device, characterized in that: include: A processor and a memory storing a computer program, wherein when the computer program is executed by the processor, the method according to any one of claims 1 to 3 is performed.
6. A computer-readable storage medium, characterized in that: Instructions are stored, and when the instructions are executed on a computer, the computer is caused to execute the method according to any one of claims 1 to 3.
Citation Information
Patent Citations
Continuous correlation machine learning system for interrupt track
CN107506444A
Interrupted track correlation method of maritime target of high-orbit optical satellite
CN110686679A
Interrupted track continuing association deep learning method
CN111898746A
Ship track fusion method and system, medium and equipment
CN114063056A
Marine target trajectory prediction method and device based on multi-source data
CN114282625A
