A method for target tracking, a target tracking device, and a drone

Abstract

The present invention is applicable to the technical field of image recognition, and provides a target tracking method, a target tracking device, and a drone. The target tracking method includes: obtaining the current position of a target human body in a current image; predicting the next position of the target human body based on the current position of the target human body; selecting a first human body to be determined in a subsequent image based on the next position; selecting a second human body to be determined in the subsequent image based on the first appearance information of the target human body, wherein the first appearance information includes human body image features; determining the position of the target human body in the subsequent image among the first human body to be determined and the second human body to be determined. In the above scheme, since the first appearance information is used to select the target human body, tracking errors due to occlusion can be avoided. In addition, the present application comprehensively determines the target human body through multiple dimensions, so it has a high tracking accuracy.

Description

Technical Field

[0001] The present invention belongs to the technical field of image recognition, and in particular relates to a target tracking method, a target tracking device and an unmanned aerial vehicle. Background Art

[0002] With the rapid development of drone-related technologies both domestically and internationally, drones are now available in a wide variety of types and applications. Drone-related functions are also becoming increasingly diverse. Among them, visual tracking is an emerging technology that has been applied to drones in recent years, often used in scenarios such as tracking pedestrians.

[0003] However, since there are often many individuals blocking the target individual during the visual tracking process of the drone, the drone often fails to track due to the switching of the tracking target. This is a technical problem that needs to be solved urgently. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a target tracking method, a target tracking device, and a drone to solve the technical problem that during the visual tracking process of the drone, there are often many individuals that block the target individual, so the drone often fails to track due to the switching of the tracking target.

[0005] A first aspect of an embodiment of the present invention provides a method for tracking a target, wherein the method includes:

[0006] Get the current position of the target body in the current image;

[0007] Predicting the next position of the target body according to the current position of the target body;

[0008] selecting a first human body to be determined in subsequent images according to the next position;

[0009] selecting a second human body to be determined in the subsequent image according to first appearance information of the target human body, wherein the first appearance information includes human body image features;

[0010] The target human body is determined among the first human body to be determined and the second human body to be determined.

[0011] Furthermore, the step of obtaining the current position of the target human body in the current image includes:

[0012] Identifying all human bodies to be identified in the current image;

[0013] The human body located at the center of the picture or the human body with the largest image area among all the human bodies to be identified is taken as the target human body, and the current position of the target human body is obtained.

[0014] Furthermore, the step of predicting the next position of the target body according to the current position of the target body includes:

[0015] The next position of the target human body is predicted according to the current position and human body data of the target human body, wherein the human body data includes human body height and human body width.

[0016] Furthermore, the step of predicting the next position of the target body according to the current position and body data of the target body includes:

[0017] Obtaining the coordinate value of the current position of the target human body and the human body data;

[0018] Calculating covariance according to the coordinate values and the human body data;

[0019] The next position of the target human body is predicted according to the covariance.

[0020] Furthermore, the step of selecting a first human body to be determined in a subsequent image according to the next position includes:

[0021] Calculating a first similarity between a first position corresponding to all human bodies in the subsequent image and the next position;

[0022] If the first similarity is higher than a first threshold, the first candidate human body corresponding to the first similarity is determined as the first human body to be determined.

[0023] Furthermore, the step of selecting a second human body to be determined in the subsequent image based on the first appearance information of the target human body includes:

[0024] Acquiring second appearance information of all human bodies in the subsequent image;

[0025] If a second similarity between the second appearance information of a second candidate human body among all the human bodies and the first appearance information is higher than a second threshold, the second candidate human body is determined to be the second human body to be determined.

[0026] Furthermore, the step of determining the position of the target human body in the subsequent image in the first to-be-determined human body and the second to-be-determined human body includes:

[0027] If the first similarity is equal to the second similarity, randomly selecting one of the first to-be-determined human body and the second to-be-determined human body as the target human body;

[0028] If the first human body to be determined and the second human body to be determined are different human bodies, and both the first similarity and the second similarity are higher than a third threshold, taking the second human body to be determined as the target human body;

[0029] If the first human body to be determined and the second human body to be determined are different human bodies, and the first similarity and the second similarity are both lower than the third threshold, the first human body to be determined is used as the target human body.

[0030] A second aspect of an embodiment of the present invention provides a target tracking device, including:

[0031] An acquisition unit, used to acquire the current position and body data of the target human body in the current image;

[0032] A prediction unit, configured to predict a next position of the target body according to the current position of the target body;

[0033] a first selecting unit, configured to select a first human body to be determined in a subsequent image according to the next position;

[0034] a second selecting unit, configured to select a second human body to be determined in the subsequent image according to first appearance information of the target human body, wherein the first appearance information includes human body image features;

[0035] A determination unit is configured to determine a position of the target human body in the subsequent image among the first human body to be determined and the second human body to be determined.

[0036] A third aspect of an embodiment of the present invention provides a drone, comprising a camera module, a motor, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method described in the first aspect when executing the computer program.

[0037] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0038] Compared to the prior art, the embodiments of the present invention have the following advantages: the present invention obtains the current position of a target person in a current image; predicts the next position of the target person based on the current position of the target person; selects a first person to be determined in a subsequent image based on the next position; selects a second person to be determined in the subsequent image based on first appearance information of the target person, wherein the first appearance information includes body image features; and determines the position of the target person in the subsequent image based on the first and second persons to be determined. The above scheme determines the target person in the subsequent image in multiple dimensions. On the one hand, the first person to be determined corresponding to the target person in the subsequent image is predicted; on the other hand, the second person to be determined corresponding to the target person in the subsequent image is selected based on the first appearance information, and the target person is determined based on the first and second persons to be determined. Because the first appearance information is used to select the target person, tracking errors due to occlusion can be avoided. Furthermore, the present invention achieves high tracking accuracy by comprehensively determining the target person across multiple dimensions. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0040] Figure 1 A schematic flow chart of a target tracking method provided by the present invention is shown;

[0041] Figure 2 A specific schematic flow chart of step 101 in a target tracking method provided by the present invention is shown;

[0042] Figure 3 A specific schematic flow chart of step 102 in a target tracking method provided by the present invention is shown;

[0043] Figure 4 A specific schematic flow chart of step 103 in a target tracking method provided by the present invention is shown;

[0044] Figure 5 FIG1 shows a specific schematic flow chart of step 104 in a target tracking method provided by the present invention;

[0045] Figure 6 A specific schematic flow chart of step 105 in a target tracking method provided by the present invention is shown.

[0046] Figure 7 A schematic diagram of a target tracking device provided by the present invention is shown;

[0047] Figure 8 FIG. 4 is a schematic diagram of a drone provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0048] In the following description, specific details such as particular system structures and techniques are provided for purposes of illustration, not limitation, to facilitate a thorough understanding of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.

[0049] It is worth noting that the target tracking method provided by the present invention can be applied to various application scenarios, such as human body tracking in video data collected by a network camera module, real-time visual tracking of drones, and other application scenarios requiring target tracking. The present invention does not impose any limitations on the application scenarios of the target tracking method. To better explain the technical solution of the present invention, the present invention uses the real-time visual tracking of drones as an example to illustrate the technical solution of the present invention.

[0050] First, the present invention provides a method for target tracking. Figure 1 , Figure 1 FIG. 1 shows a schematic flow chart of a target tracking method provided by the present invention. Figure 1 As shown, the target tracking method may include the following steps:

[0051] Step 101: Obtain the current position of the target human body in the current image.

[0052] The current image refers to the image data currently being captured by the drone. This can be the first frame captured by the drone when initializing its tracking function. The target person is the person to be tracked. The selection of the target person can be manual or automatic. When manual selection is used, the user can view the real-time image transmitted by the drone on the controller and select the corresponding person on the controller screen as the target person. When automatic selection is used, the specific execution process is as follows:

[0053] Specifically, step 101 includes steps 1011 to 1012. Figure 2 As shown, Figure 2 A specific schematic flow chart of step 101 in a target tracking method provided by the present invention is shown.

[0054] Step 1011: Identify all human bodies to be identified in the current image.

[0055] Human body recognition technology is used to identify all human bodies in the current image and obtain all human bodies to be identified.

[0056] Preferably, due to the size limitation of drones, the computing power of their processors is often small, so the present invention preferentially adopts a lightweight MobileNet-SSD model as a human body recognition method.

[0057] Step 1012: The human body located at the center of the picture or the human body with the largest image area among all the human bodies to be identified is taken as the target human body, and the current position of the target human body is obtained.

[0058] This embodiment uses two selection logics: 1. Obtain the positions of all human bodies to be identified and select the human body at the center of the image as the target human body. If there are multiple human bodies at the center of the image, the human body closest to the center of the image is selected as the target human body. 2. Since objects in the real environment generally appear larger when closer and smaller when farther away when the camera module captures images, the larger the image area occupied by a human body, the closer the human body is to the drone. Human bodies closer to the drone are often the ones that need to be tracked, so the human body with the largest image area can be selected as the target human body.

[0059] The current position of the target body is represented by coordinate values, which include the X-axis coordinate values and the Y-axis coordinate values of the upper left corner and the lower right corner of the body selection box.

[0060] Step 102: predicting the next position of the target person according to the current position of the target person.

[0061] The present invention uses a Kalman filter to predict the next position of a target body based on its current position. Kalman filtering is primarily used to track the value of a specific observed variable (coordinate value), and the basis for tracking is to predict the observed variable based on the equation of motion. For example, if the velocity of an object is known, its position at the next moment can be predicted, but this predicted value has a certain error and can only serve as a basis for tracking. Another basis is to use a measurement method to measure the value of the variable, but this predicted value also has a certain error and can only serve as a basis. Kalman filtering utilizes these two bases to perform a series of iterations to achieve the purpose of target tracking.

[0062] Kalman filtering includes two important parts: the system prediction process (predicting the system state parameters based on the system state equation) and the system measurement process (i.e., the system observation quantity, the observed state parameters and the system state parameters may be inconsistent). First, we introduce a system of discrete control processes that can be described by a linear stochastic differential equation (i.e., the system state equation):

[0063] X(k)=AX(k-1)+BU(k)+W(k)

[0064] Where A is the state transition matrix, B is the control input matrix, U(k) is the control variable of the system at time k, and W(k) is the system process noise, which is Gaussian white noise with covariance Q.

[0065] Add to this the system's measured values (i.e., observable quantities): Z(k) = HX(k) + V(k). Here, H represents the system's measurement matrix, V(k) represents the measurement noise, which is Gaussian white noise with a covariance of R, and Z(k) represents the measured value at time k, i.e., the quantity that can be observed for the system. For system state variables, not all quantities can necessarily be observed.

[0066] It is worth noting that the observation variable used in the above Kalman filter is the current position of the target body. In order to further improve the tracking accuracy, the present invention optimizes the observation variable to further improve the target tracking accuracy. The specific implementation logic is as follows:

[0067] Specifically, step 102 includes predicting the next position of the target human body according to the current position and human body data of the target human body, wherein the human body data includes human body height and human body width.

[0068] Among them, the prediction stage of the Kalman filter predicts a series of observation variables set in advance. Under normal circumstances, the observation variables generally include the target's motion speed and direction. Since the current application object is the human body, the present invention not only predicts the motion state of pedestrians, but also takes into account the target size of the human body itself. Therefore, the observation variables of the Kalman filter are increased from the original four variables x, y, dx and dy (where x and y represent the horizontal axis coordinate values, and dx and dy represent the derivatives of x and y) to eight variables x, y, w, h, dx, dy, dw, dh (where w and h represent the height and width of the human body, and dw and dh represent the derivatives of dw and dh). It can be understood that the solution using the above-mentioned observation variables can predict the height and width of the human body while predicting the motion state. In the case where the tracked target appears to be larger or smaller, misjudgment of the tracked target can be discovered in a timely manner, thereby improving the accuracy of target tracking.

[0069] Specifically, the above calculation process includes the following steps 1021 to 1023. Figure 3 As shown, Figure 3 A specific schematic flow chart of step 102 in a target tracking method provided by the present invention is shown.

[0070] Step 1021: Acquire the coordinate value of the current position of the target human body and the human body data.

[0071] The processor obtains the coordinate values of the upper left corner and the lower right corner of the selected box corresponding to the target human body. The processor obtains the human body data of the target human body, which includes but is not limited to one data such as human body height or human body width or a combination of multiple data.

[0072] Step 1022: Calculate the covariance based on the coordinate values and the human body data.

[0073] Step 1023: predict the next position of the target person based on the covariance.

[0074] The data in the next position includes but is not limited to the X-axis coordinate value, the Y-axis coordinate value, the human body height H, and the human body width W.

[0075] In this embodiment, the application scenarios of human body tracking are fully considered, so human body data is innovatively added to the observation variables in the traditional Kalman filter, which not only improves the data dimension of target tracking, but also has high adaptability to the application scenarios of human body tracking, so it has higher tracking accuracy.

[0076] Step 103: Selecting a first human body to be determined in subsequent images according to the next position.

[0077] The subsequent image is the image that follows the current image in time sequence, i.e. the current image is in front and the subsequent image is behind. Since the position of the target person changes from the current image to the subsequent image, it is necessary to track the target in the subsequent image.

[0078] Specifically, step 103 includes steps 1031 to 1032. Figure 4 As shown, Figure 4 A specific schematic flow chart of step 103 in a target tracking method provided by the present invention is shown.

[0079] Step 1031: Calculate a first similarity between the first position corresponding to all human bodies in the subsequent image and the next position.

[0080] The data at the first position includes, but is not limited to, an X-axis coordinate value, a Y-axis coordinate value, a person's height H, and a person's width W. The data at the next position includes, but is not limited to, an X-axis coordinate value, a Y-axis coordinate value, a person's height H, and a person's width W. A first similarity between the first position and the next position is calculated. The first similarity can be calculated based on the cosine theorem or the Euclidean distance.

[0081] Step 1032: If the first similarity is higher than a first threshold, the first candidate human body corresponding to the first similarity is determined as the first human body to be determined.

[0082] Step 104: selecting a second human body to be determined in the subsequent image based on the first appearance information of the target human body, wherein the first appearance information includes human body image features.

[0083] Since steps 101 to 103 only judge the motion state of the target human body, when other human bodies block the target human body, the other human bodies will be identified as the tracking target (because the motion state of the target human body and the other human bodies are consistent), which will lead to tracking errors. Therefore, in order to solve the above technical problems, the present invention combines appearance information for comprehensive judgment. Among them, the first appearance information includes but is not limited to human body image characteristics, which include one or a combination of multiple information such as clothing color, body shape (height, weight, and body shape), hairstyle, clothing style or hair color.

[0084] Specifically, step 104 includes steps 1041 to 1042. Figure 5 As shown, Figure 5 A specific schematic flow chart of step 104 in a target tracking method provided by the present invention is shown.

[0085] Step 1041: Acquire second appearance information of all human bodies in the subsequent image.

[0086] The second appearance information includes but is not limited to human body features, which include clothing color, body type (height, weight, or thin), hairstyle, clothing style or hair color, and the like, or a combination of multiple pieces of information.

[0087] The recognition of the first or second appearance information can be processed using a convolutional neural network. However, single-target tracking methods based on deep learning typically have a large network architecture, making them unsuitable for deployment on a chip. In view of this, the present invention provides a lightweight convolutional neural network for recognizing appearance information.

[0088] The lightweight convolutional neural network provided by the present invention is an improved network model based on the traditional resnet18 network. Among them, the traditional resnet18 network has 17 convolutional layers and 1 fully connected layer, a total of 18 layers. In the present invention, the first 7*7 convolutional layer in the original resnet18 network is changed to two 3*3 convolutional layers to reduce the number of model parameters and speed up the speed. There are 4 blocks under the 7*7 convolutional layer in the original resnet18 network, each block contains 4 convolutional layers, and the 4 convolutional layers are divided into two convolutional blocks, so the 4 blocks contain a total of 16 convolutional layers. The present invention reduces the convolutional blocks in each block to 1 (that is, only two convolutional layers are left), and only 8 convolutional layers are retained. The modified resnet18 network only has the above-mentioned 10-layer network structure, the model size is small, and it is easy to deploy on a drone.

[0089] Step 1042: If the second similarity between the second appearance information of the second candidate human body among all the human bodies and the first appearance information is higher than a second threshold, the second candidate human body is determined as the second human body to be determined.

[0090] Calculate the cosine similarity between the first appearance information and the second appearance information, and use the cosine similarity as the second similarity. If the second similarity exceeds a second threshold, determine the second candidate human body as the second human body to be determined. The first threshold and the second threshold can be the same or different, and are not limited here. Preferably, the first threshold and the second threshold can be set to 0.9.

[0091] In this embodiment, the target human body is tracked by matching the first appearance information of the target human body with the second appearance information of all human bodies to screen the target human body from all human bodies. Since the appearance information can effectively avoid misjudgment due to human body occlusion, the accuracy of target tracking is further improved.

[0092] Step 105: Determine the target human body among the first human bodies to be determined and the second human bodies to be determined.

[0093] The processor determines which of the two pending human bodies is the target human body based on the similarity between the first pending human body and the second pending human body. The specific judgment logic is as follows:

[0094] Specifically, step 105 includes steps 1051 to 1053. Figure 6 As shown, Figure 6 A specific schematic flow chart of step 105 in a target tracking method provided by the present invention is shown.

[0095] Step 1051: If the first similarity is equal to the second similarity, randomly selecting one of the first to-be-determined human body and the second to-be-determined human body as the target human body.

[0096] For example, if the first similarity and the second similarity are both 0.9, the first human body to be determined or the second human body to be determined is used as the target human body.

[0097] Step 1052: If the first human body to be determined and the second human body to be determined are different human bodies, and the first similarity and the second similarity are both higher than a third threshold, the second human body to be determined is used as the target human body.

[0098] It can be understood that the tracking method of human body prediction is adopted in steps 101 to 103, and the tracking method of appearance information matching is adopted in step 104. If the first human body to be determined and the second human body to be determined are different human bodies, it means that the results of the two algorithms are different. If the first similarity and the second similarity are both higher than the third threshold, the second human body to be determined is used as the target human body (that is, the method adopted in step 104 is used as the basis). For example: if the first similarity and the second similarity are both higher than 0.9, the second human body to be determined is used as the target human body.

[0099] Step 1053: If the first human body to be determined and the second human body to be determined are different human bodies, and the first similarity and the second similarity are both lower than the third threshold, the first human body to be determined is used as the target human body.

[0100] If the first person to be determined and the second person to be determined are different people, it means that the results of the two algorithms are different. If the first similarity and the second similarity are both lower than the third threshold, the first person to be determined is used as the target person (i.e., the method used in steps 101 to 103 is used as the basis). For example: if the first similarity and the second similarity are both lower than 0.9, the first person to be determined is used as the target person.

[0101] As an embodiment of the present invention, after determining the target human body in the subsequent image, the drone is controlled to move according to the position change of the target human body to achieve the function of drone following.

[0102] In this embodiment, the current position of a target person in a current image is obtained; based on the current position of the target person, the next position of the target person is predicted; based on the next position, a first person to be determined is selected in a subsequent image; based on first appearance information of the target person, a second person to be determined is selected in the subsequent image, wherein the first appearance information includes body image features; and the position of the target person in the subsequent image is determined between the first and second persons to be determined. The above scheme determines the target person in the subsequent image in multiple dimensions. On the one hand, the first person to be determined corresponding to the target person in the subsequent image is predicted; on the other hand, the second person to be determined corresponding to the target person in the subsequent image is selected based on the first appearance information, and the target person is determined between the first and second persons to be determined. Because the first appearance information is used to select the target person, tracking errors due to occlusion can be avoided. Furthermore, the present application comprehensively determines the target person through multiple dimensions, thus achieving high tracking accuracy.

[0103] like Figure 7 The present invention provides a target tracking device 7, see Figure 7 , Figure 7 A schematic diagram of a target tracking device provided by the present invention is shown in FIG. Figure 7 The target tracking device includes:

[0104] An acquisition unit 71 is used to acquire the current position and body data of the target human body in the current image;

[0105] A prediction unit 73 is configured to predict the next position of the target body according to the current position of the target body;

[0106] A first selecting unit 73 is configured to select a first human body to be determined in subsequent images according to the next position;

[0107] A second selecting unit 74 is configured to select a second human body to be determined in the subsequent image based on first appearance information of the target human body, wherein the first appearance information includes human body image features;

[0108] The determining unit 75 is configured to determine a position of the target human body in the first to-be-determined human body and the second to-be-determined human body in the subsequent image.

[0109] The present invention provides a target tracking device that obtains the current position of a target person in a current image; predicts the next position of the target person based on the current position of the target person; selects a first person to be determined in a subsequent image based on the next position; selects a second person to be determined in the subsequent image based on first appearance information of the target person, wherein the first appearance information includes body image features; and determines the position of the target person in the subsequent image based on the first and second persons to be determined. The above scheme determines the target person in the subsequent image in multiple dimensions. On the one hand, the first person to be determined corresponding to the target person in the subsequent image is predicted; on the other hand, the second person to be determined corresponding to the target person in the subsequent image is selected based on the first appearance information, and the target person is determined based on the first and second persons to be determined. Because the first appearance information is used to select the target person, tracking errors due to occlusion can be avoided. Furthermore, the present application comprehensively determines the target person through multiple dimensions, thus achieving high tracking accuracy.

[0110] Figure 8 FIG is a schematic diagram of a drone provided by an embodiment of the present invention. Figure 8 As shown, a drone 8 of this embodiment includes: a camera module 80, a motor 81, a processor 82, a memory 83, and a computer program 84 stored in the memory 83 and executable on the processor 82, such as a target tracking method program. When the processor 82 executes the computer program 84, the steps of each of the above-mentioned target tracking method embodiments are implemented, such as Figure 1 Alternatively, when the processor 82 executes the computer program 84, the functions of each unit in the above-mentioned device embodiments are realized, for example, Figure 7 The functions of the units 71 to 75 are shown.

[0111] Exemplarily, the computer program 84 may be divided into one or more units, which are stored in the memory 83 and executed by the processor 82 to implement the present invention. The one or more units may be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program 84 in the drone 8. For example, the specific functions of each unit of the computer program 84 may be as follows:

[0112] An acquisition unit, used to acquire the current position and body data of the target human body in the current image;

[0113] A prediction unit, configured to predict a next position of the target body according to the current position of the target body;

[0114] a first selecting unit, configured to select a first human body to be determined in a subsequent image according to the next position;

[0115] a second selecting unit, configured to select a second human body to be determined in the subsequent image according to first appearance information of the target human body, wherein the first appearance information includes human body image features;

[0116] A determination unit is configured to determine a position of the target human body in the subsequent image among the first human body to be determined and the second human body to be determined.

[0117] The drone includes but is not limited to a processor 82 and a memory 83. Those skilled in the art will appreciate that Figure 8 This is merely an example of a drone 8 and does not constitute a limitation on the drone 8. The drone 8 may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the drone may also include input and output devices, network access devices, buses, etc.

[0118] The processor 82 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0119] The memory 83 may be an internal storage unit of the drone 8, such as a hard drive or memory of the drone 8. The memory 83 may also be an external storage device of the drone 8, such as a plug-in hard drive, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the drone 8. Furthermore, the memory 83 may include both an internal storage unit of the drone 8 and an external storage device. The memory 83 is used to store the computer program and other programs and data required by the roaming control device. The memory 83 may also be used to temporarily store data that has been output or is about to be output.

[0120] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0121] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the embodiment of the method of the present invention. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0122] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0123] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned various method embodiments can be implemented.

[0124] An embodiment of the present invention provides a computer program product. When the computer program product is run on a mobile terminal, the mobile terminal can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0125] If the integrated unit is implemented as a software functional unit and sold or used as a standalone product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the process steps in the above-mentioned method embodiments by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include at least: any entity or device capable of carrying the computer program code to the camera / drone, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium. For example, a USB flash drive, mobile hard drive, magnetic disk, or optical disk. In some jurisdictions, based on legislation and patent practice, computer-readable media cannot be electric carrier signals or telecommunication signals.

[0126] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0127] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0128] In the embodiments provided by the present invention, it should be understood that the disclosed devices / network equipment and methods can be implemented in other ways. For example, the device / network equipment embodiments described above are merely illustrative. For example, the division of the modules or units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0129] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units.

[0130] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0131] It will also be understood that the term "and / or" used in the present description and appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0132] As used in the present specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to monitoring," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is monitored" may be interpreted as meaning "upon determination" or "in response to determining" or "upon monitoring [described condition or event]" or "in response to monitoring [described condition or event]," depending on the context.

[0133] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0134] References to "one embodiment" or "some embodiments" in the present specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present invention. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in yet other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "including," "comprising," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0135] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A target tracking method, characterized in that: The method comprises: Get the current position of the target body in the current image; Predicting the next position of the target body through Kalman filtering according to the current position of the target body; wherein the Kalman filtering includes eight observation variables; the eight observation variables include a horizontal axis coordinate value, a vertical axis coordinate value, a body height, a body width, a derivative of the horizontal axis coordinate value, a derivative of the vertical axis coordinate value, a derivative of the body height, and a derivative of the body width; Calculating a first similarity between a first position corresponding to all human bodies in subsequent images and the next position; If the first similarity is higher than a first threshold, determining the first candidate human body corresponding to the first similarity as the first human body to be determined; Acquiring second appearance information of all human bodies in the subsequent image; If a second similarity between the second appearance information of a second candidate human body among all the human bodies and the first appearance information is higher than a second threshold, determining the second candidate human body as the second human body to be determined, wherein the first appearance information includes human body image features; If the first similarity is equal to the second similarity, randomly selecting one of the first to-be-determined human body and the second to-be-determined human body as the target human body; If the first human body to be determined and the second human body to be determined are different human bodies, and both the first similarity and the second similarity are higher than a third threshold, taking the second human body to be determined as the target human body; If the first human body to be determined and the second human body to be determined are different human bodies, and the first similarity and the second similarity are both lower than the third threshold, the first human body to be determined is used as the target human body.

2. The target tracking method according to claim 1, wherein: The step of obtaining the current position of the target human body in the current image includes: Identifying all human bodies to be identified in the current image; The human body located at the center of the picture or the human body with the largest image area among all the human bodies to be identified is taken as the target human body, and the current position of the target human body is obtained.

3. The target tracking method according to claim 1, wherein: The step of predicting the next position of the target body according to the current position of the target body comprises: The next position of the target human body is predicted according to the current position and human body data of the target human body, wherein the human body data includes human body height and human body width.

4. The target tracking method according to claim 3, wherein: The step of predicting the next position of the target body according to the current position and body data of the target body comprises: Obtaining the coordinate value of the current position of the target human body and the human body data; Calculating covariance according to the coordinate values and the human body data; The next position of the target human body is predicted according to the covariance.

5. A target tracking device, characterized in that: The device comprises: An acquisition unit, used to acquire the current position and body data of the target human body in the current image; A prediction unit, configured to predict the next position of the target person through a Kalman filter based on the current position of the target person; wherein the Kalman filter includes eight observation variables; the eight observation variables include a horizontal axis coordinate value, a vertical axis coordinate value, a person's height, a person's width, a derivative of the horizontal axis coordinate value, a derivative of the vertical axis coordinate value, a derivative of the person's height, and a derivative of the person's width; A first selection unit is used to calculate a first similarity between a first position corresponding to all human bodies in subsequent images and the next position; If the first similarity is higher than a first threshold, determining the first candidate human body corresponding to the first similarity as the first human body to be determined; a second selecting unit, configured to obtain second appearance information of all human bodies in the subsequent image; If a second similarity between the second appearance information of a second candidate human body among all the human bodies and the first appearance information is higher than a second threshold, determining the second candidate human body as the second human body to be determined, wherein the first appearance information includes human body image features; a determining unit, configured to randomly select one of the first to-be-determined human body and the second to-be-determined human body as the target human body if the first similarity is equal to the second similarity; If the first human body to be determined and the second human body to be determined are different human bodies, and both the first similarity and the second similarity are higher than a third threshold, taking the second human body to be determined as the target human body; If the first human body to be determined and the second human body to be determined are different human bodies, and the first similarity and the second similarity are both lower than the third threshold, the first human body to be determined is used as the target human body.

6. A drone comprising a camera module, a motor, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images