A method and system for recognizing hand gestures for helicopter lifeguard search and rescue operations

Abstract

The present invention discloses a method and system for identifying gestures for search and rescue operations of helicopter lifeguards, and relates to the technical field of rescue gesture recognition. The method comprises: obtaining a training data set, wherein the training data set comprises: a feature vector and a gesture label of a gesture, each of the gesture labels corresponding to an operation category, wherein the feature vector of each gesture in the training data set corresponds to an operation category of a gesture label; setting a gesture label prediction model to which the gesture belongs, obtaining the feature vector of the current gesture, and obtaining the operation category of the rescue operation to which the feature vector #imgabs0# of the current gesture belongs based on the training data set; optimizing the gesture label prediction model to which the gesture belongs, and obtaining the feature vector of the gesture during the search and rescue operation of the helicopter lifeguard in real time, inputting the feature vector into the optimized gesture label prediction model, and identifying the gesture during the search and rescue operation of the helicopter lifeguard according to the operation category.

Description

Technical Field

[0001] The present invention belongs to the technical field of rescue gesture recognition, and more specifically, relates to a method and system for recognizing gestures of helicopter lifeguard search and rescue operations. Background Art

[0002] Helicopter rescuers often use hand signals to communicate and direct ground personnel or other rescuers during search and rescue operations. The following are some of the hand signals that may be used:

[0003] Smooth landing: Raise your hands horizontally and move them downward to indicate that you are about to land and that you need to land smoothly.

[0004] Emergency: Crossing your hands above your head indicates an emergency situation that requires immediate action or cessation of an operation.

[0005] Moving Forward: One hand extended forward with the palm facing downward indicates the need to move forward or in the direction of travel.

[0006] Stop: Extend your hands forward with your palms facing outward to indicate that you need to stop the current action or operation.

[0007] Waiting for instructions: Crossing your hands in front of your chest indicates that you need to wait for instructions or commands.

[0008] Warning: A hand raised upward with the palm facing outward, indicating a need for attention or warning to those around you.

[0009] Danger: The finger points to the ground and makes a magnified gesture to indicate danger or the need for special attention.

[0010] Stretcher: Make pulling motions with both hands to simulate the shape of a stretcher, indicating that a stretcher is needed or rescue operations are being carried out.

[0011] Currently, helicopter pilots generally observe with their eyes. However, due to the complex search and rescue environment, observation with the human eye alone may lead to misjudgment. Therefore, there is an urgent need for a technical solution that can automatically and efficiently recognize search and rescue gestures and provide feedback to pilots and other operators. Summary of the Invention

[0012] To solve the above technical problems, the present invention proposes a method for recognizing hand gestures of helicopter lifeguard search and rescue operations, comprising:

[0013] Acquire a training data set, the training data set comprising: a feature vector of a gesture and a gesture label, each of the gesture labels corresponding to a task category, wherein the feature vector of each gesture in the training data set corresponds to a task category of a gesture label;

[0014] Setting a gesture label prediction model to which the gesture belongs, obtaining a feature vector of the current gesture, and obtaining, based on the training data set, an operation category of the rescue operation to which the feature vector of the current gesture belongs;

[0015] The gesture label prediction model to which the gesture belongs is optimized, and the feature vector of the gesture during the helicopter lifeguard search and rescue operation is obtained in real time and input into the optimized gesture label prediction model to identify the gesture during the helicopter lifeguard search and rescue operation according to the operation category.

[0016] Furthermore, the gesture label prediction model to which the gesture belongs includes:

[0017] ,

[0018] in, is the feature vector of the current gesture The type of rescue operation to which it belongs, is the number of samples in the training dataset, The training data set The feature vector of the gesture In the Lagrange multipliers on the job categories, The training data set The feature vector of the gesture Corresponding to The gesture label of the job category, if the training dataset The feature vector of the gesture Belong to job categories, then , otherwise 0, is the kernel function used to calculate the feature vector of the current gesture Compared with the training data set The feature vector of the gesture The similarity of For the The offset item for each job category.

[0019] Furthermore, the kernel function include:

[0020] ,

[0021] in, is the adjustment factor, For the The weight of the dimensions, is the dimension of the gesture’s feature vector, The current gesture is in The feature vector of dimension The training data set The gesture in A feature vector of dimensions.

[0022] Furthermore, optimizing the gesture label prediction model of the gesture includes: calculating the first The feature vector of the gesture In the Lagrange multipliers on job categories Hedi Offset items for job categories , specifically:

[0023] ,

[0024] in, is the number of job categories, The training data set The feature vector of the gesture In the Lagrange multipliers on the job categories, The training data set The feature vector of the gesture Corresponding to Gesture labels for each task category.

[0025] Furthermore, set constraints, specifically:

[0026] ,

[0027] ,

[0028] in, is the regularization parameter used to control the calculation and The complexity of time.

[0029] The present invention also proposes a helicopter lifeguard search and rescue operation gesture recognition system, comprising:

[0030] A data acquisition module is used to acquire a training data set, wherein the training data set includes: a feature vector of a gesture and a gesture label, each of the gesture labels corresponds to a job category, wherein the feature vector of each gesture in the training data set corresponds to a job category of a gesture label;

[0031] A training model module is used to set a gesture label prediction model to which a gesture belongs, obtain a feature vector of a current gesture, and obtain, based on the training data set, an operation category of a rescue operation to which the feature vector of the current gesture belongs;

[0032] The recognition module is used to optimize the gesture label prediction model to which the gesture belongs, and to obtain the feature vector of the gesture during the helicopter lifeguard search and rescue operation in real time, and input it into the optimized gesture label prediction model to identify the gesture during the helicopter lifeguard search and rescue operation according to the operation category.

[0033] Furthermore, the gesture label prediction model to which the gesture belongs includes:

[0034] ,

[0035] in, is the feature vector of the current gesture The type of rescue operation to which it belongs, is the number of samples in the training dataset, The training data set The feature vector of the gesture In the Lagrange multipliers on the job categories, The training data set The feature vector of the gesture Corresponding to The gesture label of the job category, if the training dataset The feature vector of the gesture Belong to job categories, then , otherwise 0, is the kernel function used to calculate the feature vector of the current gesture Compared with the training data set The feature vector of the gesture The similarity of For the The offset item for each job category.

[0036] Furthermore, the kernel function include:

[0037] ,

[0038] in, is the adjustment factor, For the The weight of the dimensions, is the dimension of the gesture’s feature vector, The current gesture is in The feature vector of dimension The training data set The gesture in A feature vector of dimensions.

[0039] Furthermore, optimizing the gesture label prediction model of the gesture includes: calculating the first The feature vector of the gesture In the Lagrange multipliers on job categories Hedi Offset items for job categories , specifically:

[0040] ,

[0041] in, is the number of job categories, The training data set The feature vector of the gesture In the Lagrange multipliers on the job categories, The training data set The feature vector of the gesture Corresponding to Gesture labels for each task category.

[0042] Furthermore, set constraints, specifically:

[0043] ,

[0044] ,

[0045] in, is the regularization parameter used to control the calculation and The complexity of time.

[0046] Compared with the prior art, the above technical solution conceived by the present invention has the following beneficial effects:

[0047] The present invention obtains a training data set, which includes: a feature vector and a gesture label of a gesture, each of which corresponds to a job category, wherein the feature vector of each gesture in the training data set corresponds to a job category of a gesture label; sets a gesture label prediction model to which the gesture belongs, obtains the feature vector of the current gesture, and obtains the job category of the rescue operation to which the feature vector of the current gesture belongs based on the training data set; optimizes the gesture label prediction model to which the gesture belongs, and obtains the feature vector of the gesture during the helicopter lifeguard search and rescue operation in real time, inputs it into the optimized gesture label prediction model to which the gesture belongs, and recognizes the gesture of the helicopter lifeguard during the search and rescue operation according to the job category. Through the above technical solution, the present invention can automatically recognize the gestures of the helicopter lifeguard during the search and rescue operation, especially the gestures made by the lifeguard on the ground to the helicopter, so as to complete the corresponding rescue action according to the job category to which the gesture belongs. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a flow chart of the method of embodiment 1 of the present invention;

[0049] Figure 2 This is a system structure diagram of Example 2 of the present invention. DETAILED DESCRIPTION

[0050] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0051] The method provided by the present invention can be implemented in the following terminal environment, wherein the terminal may include one or more of the following components: a processor, a storage medium, and a display screen. The storage medium stores at least one instruction, which is loaded and executed by the processor to implement the method described in the following embodiments.

[0052] A processor can include one or more processing cores. It connects various components within the terminal using various interfaces and circuits. It executes instructions, programs, code sets, or instruction sets stored in storage media, and accesses data stored in storage media to perform various terminal functions and process data.

[0053] The storage medium may include a random access memory (RAM) or a read-only memory (ROM). The storage medium may be used to store instructions, programs, codes, code sets, or instructions.

[0054] The display is used to show the interactive sections of each application.

[0055] All subscripts in the formulas of the present invention are only used to distinguish parameters and have no actual meaning.

[0056] In addition, those skilled in the art will appreciate that the structure of the terminal described above does not limit the terminal. The terminal may include more or fewer components, or a combination of certain components, or a different arrangement of components. For example, the terminal may also include a radio frequency circuit, an input unit, a sensor, an audio circuit, a power supply, and other components, which will not be described in detail here.

[0057] Example 1

[0058] like Figure 1 As shown, an embodiment of the present invention provides a method for recognizing hand gestures for a helicopter lifeguard search and rescue operation, comprising:

[0059] Step 101: Acquire a training data set, wherein the training data set includes: a feature vector of a gesture and a gesture label, each of the gesture labels corresponds to a task category, wherein the feature vector of each gesture in the training data set corresponds to a task category of a gesture label;

[0060] The feature vector of a gesture generally includes:

[0061] Hand position: The position coordinates of the hand on the two-dimensional plane.

[0062] Gesture shape: The shape characteristics of the hand can be represented by geometric features such as the curvature of the hand contour.

[0063] Motion trajectory: The motion trajectory of the hand over a period of time can be represented by the change of position coordinates.

[0064] Finger posture: The relative position and angle of the fingers can be expressed by the position of the joints of the hand.

[0065] Operation categories generally include: descending, ascending, hovering, etc.

[0066] Step 102: Setting a gesture label prediction model to which the gesture belongs, obtaining a feature vector of the current gesture, and obtaining the operation category of the rescue operation to which the feature vector of the current gesture belongs based on the training data set;

[0067] Specifically, the gesture label prediction model of the gesture includes:

[0068] ,

[0069] in, is the feature vector of the current gesture The type of rescue operation to which it belongs, is the number of samples in the training dataset, The training data set The feature vector of the gesture In the Lagrange multipliers on the job categories, The training data set The feature vector of the gesture Corresponding to The gesture label of the job category, if the training dataset The feature vector of the gesture Belong to job categories, then , otherwise 0, is the kernel function used to calculate the feature vector of the current gesture Compared with the training data set The feature vector of the gesture The similarity of For the The offset item for each job category.

[0070] Specifically, the kernel function include:

[0071] ,

[0072] in, is the adjustment factor, For the The weight of the dimensions, is the dimension of the gesture’s feature vector, The current gesture is in The feature vector of dimension The training data set The gesture in A feature vector of dimensions.

[0073] Step 103: Optimize the gesture label prediction model to which the gesture belongs, obtain the feature vector of the gesture during the helicopter lifeguard search and rescue operation in real time, input it into the optimized gesture label prediction model to identify the gesture during the helicopter lifeguard search and rescue operation according to the operation category.

[0074] Specifically, optimizing the gesture label prediction model of the gesture includes: calculating the first The feature vector of the gesture In the Lagrange multipliers on job categories Hedi Offset items for job categories , specifically:

[0075] ,

[0076] in, is the number of job categories, The training data set The feature vector of the gesture In the Lagrange multipliers on the job categories, The training data set The feature vector of the gesture Corresponding to Gesture labels for each task category.

[0077] Specifically, set constraints, specifically:

[0078] ,

[0079] ,

[0080] in, is the regularization parameter used to control the calculation and The complexity of time.

[0081] Example 2

[0082] like Figure 2 As shown, an embodiment of the present invention further provides a helicopter lifeguard search and rescue operation gesture recognition system, comprising:

[0083] A data acquisition module is used to acquire a training data set, wherein the training data set includes: a feature vector of a gesture and a gesture label, each of the gesture labels corresponds to a job category, wherein the feature vector of each gesture in the training data set corresponds to a job category of a gesture label;

[0084] A training model module is used to set a gesture label prediction model to which a gesture belongs, obtain a feature vector of a current gesture, and obtain, based on the training data set, an operation category of a rescue operation to which the feature vector of the current gesture belongs;

[0085] Specifically, the gesture label prediction model of the gesture includes:

[0086] ,

[0087] in, is the feature vector of the current gesture The type of rescue operation to which it belongs, is the number of samples in the training dataset, The training data set The feature vector of the gesture In the Lagrange multipliers on the job categories, The training data set The feature vector of the gesture Corresponding to The gesture label of the job category, if the training dataset The feature vector of the gesture Belong to job categories, then , otherwise 0, is the kernel function used to calculate the feature vector of the current gesture Compared with the training data set The feature vector of the gesture The similarity of For the The offset item for each job category.

[0088] Specifically, the kernel function include:

[0089] ,

[0090] in, is the adjustment factor, For the The weight of the dimensions, is the dimension of the gesture’s feature vector, The current gesture is in The feature vector of dimension The training data set The gesture in A feature vector of dimensions.

[0091] The recognition module is used to optimize the gesture label prediction model to which the gesture belongs, and to obtain the feature vector of the gesture during the helicopter lifeguard search and rescue operation in real time, and input it into the optimized gesture label prediction model to identify the gesture during the helicopter lifeguard search and rescue operation according to the operation category.

[0092] Specifically, optimizing the gesture label prediction model of the gesture includes: calculating the first The feature vector of the gesture In the Lagrange multipliers on job categories Hedi Offset items for job categories , specifically:

[0093] ,

[0094] in, is the number of job categories, The training data set The feature vector of the gesture In the Lagrange multipliers on the job categories, The training data set The feature vector of the gesture Corresponding to Gesture labels for each task category.

[0095] Specifically, set constraints, specifically:

[0096] ,

[0097] ,

[0098] in, is the regularization parameter used to control the calculation and The complexity of time.

[0099] Example 3

[0100] An embodiment of the present invention further provides a storage medium storing a plurality of instructions, wherein the instructions are used to implement the aforementioned method for recognizing gestures for helicopter lifeguard search and rescue operations.

[0101] Optionally, in this embodiment, the storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0102] Optionally, in this embodiment, the storage medium is configured to store program codes for executing the following steps: Step 101, obtaining a training data set, the training data set comprising: a feature vector of a gesture and a gesture label, each of the gesture labels corresponding to a job category, wherein the feature vector of each gesture in the training data set corresponds to a job category of a gesture label;

[0103] Step 102: Setting a gesture label prediction model to which the gesture belongs, obtaining a feature vector of the current gesture, and obtaining the operation category of the rescue operation to which the feature vector of the current gesture belongs based on the training data set;

[0104] Specifically, the gesture label prediction model of the gesture includes:

[0105] ,

[0106] in, is the feature vector of the current gesture The type of rescue operation to which it belongs, is the number of samples in the training dataset, The training data set The feature vector of the gesture In the Lagrange multipliers on the job categories, The training data set The feature vector of the gesture Corresponding to The gesture label of the job category, if the training dataset The feature vector of the gesture Belong to job categories, then , otherwise 0, is the kernel function used to calculate the feature vector of the current gesture Compared with the training data set The feature vector of the gesture The similarity of For the The offset item for each job category.

[0107] Specifically, the kernel function include:

[0108] ,

[0109] in, is the adjustment factor, For the The weight of the dimensions, is the dimension of the gesture’s feature vector, The current gesture is in The feature vector of dimension The training data set The gesture in A feature vector of dimensions.

[0110] Step 103: Optimize the gesture label prediction model to which the gesture belongs, obtain the feature vector of the gesture during the helicopter lifeguard search and rescue operation in real time, input it into the optimized gesture label prediction model to identify the gesture during the helicopter lifeguard search and rescue operation according to the operation category.

[0111] Specifically, optimizing the gesture label prediction model of the gesture includes: calculating the first The feature vector of the gesture In the Lagrange multipliers on job categories Hedi Offset items for job categories , specifically:

[0112] ,

[0113] in, is the number of job categories, The training data set The feature vector of the gesture In the Lagrange multipliers on the job categories, The training data set The feature vector of the gesture Corresponding to Gesture labels for each task category.

[0114] Specifically, set constraints, specifically:

[0115] ,

[0116] ,

[0117] in, is the regularization parameter used to control the calculation and The complexity of time.

[0118] Example 4

[0119] An embodiment of the present invention also proposes an electronic device, including a processor and a storage medium connected to the processor, wherein the storage medium stores multiple instructions, which can be loaded and executed by the processor to enable the processor to execute a gesture recognition method for helicopter lifeguard search and rescue operations.

[0120] Specifically, the electronic device of this embodiment may be a computer terminal, which may include: one or more processors and a storage medium.

[0121] The storage medium can be used to store software programs and modules, such as the helicopter lifeguard search and rescue operation gesture recognition method in an embodiment of the present invention, and the corresponding program instructions / modules. The processor executes the software programs and modules stored in the storage medium to perform various functional applications and data processing, thereby implementing the aforementioned helicopter lifeguard search and rescue operation gesture recognition method. The storage medium may include high-speed random access memory and may also include non-volatile storage media, such as one or more magnetic storage systems, flash memory, or other non-volatile solid-state storage media. In some instances, the storage medium may further include storage media remotely located relative to the processor, and these remote storage media may be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0122] The processor can call the information and application stored in the storage medium through the transmission system to execute the following steps: Step 101, obtaining a training data set, wherein the training data set includes: a feature vector and a gesture label of a gesture, each of the gesture labels corresponding to a job category, wherein the feature vector of each gesture in the training data set corresponds to a job category of a gesture label;

[0123] Step 102: Setting a gesture label prediction model to which the gesture belongs, obtaining a feature vector of the current gesture, and obtaining the operation category of the rescue operation to which the feature vector of the current gesture belongs based on the training data set;

[0124] Specifically, the gesture label prediction model of the gesture includes:

[0125] ,

[0126] in, is the feature vector of the current gesture The type of rescue operation to which it belongs, is the number of samples in the training dataset, The training data set The feature vector of the gesture In the Lagrange multipliers on the job categories, The training data set The feature vector of the gesture Corresponding to The gesture label of the job category, if the training dataset The feature vector of the gesture Belong to job categories, then , otherwise 0, is the kernel function used to calculate the feature vector of the current gesture Compared with the training data set The feature vector of the gesture The similarity of For the The offset item for each job category.

[0127] Specifically, the kernel function include:

[0128] ,

[0129] in, is the adjustment factor, For the The weight of the dimensions, is the dimension of the gesture’s feature vector, The current gesture is in The feature vector of dimension The training data set The gesture in A feature vector of dimensions.

[0130] Step 103: Optimize the gesture label prediction model to which the gesture belongs, obtain the feature vector of the gesture during the helicopter lifeguard search and rescue operation in real time, input it into the optimized gesture label prediction model to identify the gesture during the helicopter lifeguard search and rescue operation according to the operation category.

[0131] Specifically, optimizing the gesture label prediction model of the gesture includes: calculating the first The feature vector of the gesture In the Lagrange multipliers on job categories Hedi Offset items for job categories , specifically:

[0132] ,

[0133] in, is the number of job categories, The training data set The feature vector of the gesture In the Lagrange multipliers on the job categories, The training data set The feature vector of the gesture Corresponding to Gesture labels for each task category.

[0134] Specifically, set constraints, specifically:

[0135] ,

[0136] ,

[0137] in, is the regularization parameter used to control the calculation and The complexity of time.

[0138] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0139] In the above embodiments of the present invention, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0140] In the several embodiments provided by the present invention, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the system embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, and can be electrical or other forms.

[0141] 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, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0142] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0143] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only storage medium (ROM, Read-Only Memory), random access storage medium (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, and other media that can store program code.

[0144] Obviously, the above embodiments are merely examples for clarity of explanation and are not intended to limit the implementation methods. Those skilled in the art will readily appreciate that other variations or modifications based on the above descriptions are possible. It is not necessary and impossible to enumerate all implementation methods here. Obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A method for recognizing hand gestures for helicopter lifeguard search and rescue operations, characterized in that: include: Acquire a training data set, the training data set comprising: a feature vector of a gesture and a gesture label, each of the gesture labels corresponding to a task category, wherein the feature vector of each gesture in the training data set corresponds to a task category of a gesture label; Setting a gesture label prediction model to which the gesture belongs, obtaining a feature vector of the current gesture, and obtaining, based on the training data set, an operation category of the rescue operation to which the feature vector of the current gesture belongs; The gesture label prediction model to which the gesture belongs includes: Where f(x′) is the category of the rescue operation to which the feature vector x′ of the current gesture belongs, N is the number of samples in the training dataset, and α i，c is the feature vector x of the i-th gesture in the training dataset i Lagrange multiplier on the cth job category, y i，c is the feature vector x of the i-th gesture in the training dataset i The gesture label corresponding to the c-th job category is, if the feature vector x of the i-th gesture in the training dataset is i Belongs to the cth job category, then y i，c =1, otherwise 0, K(x′, x i ) is a kernel function used to calculate the feature vector x′ of the current gesture and the feature vector x of the i-th gesture in the training data set. i The similarity of b c is the bias term for the c-th job category; The kernel function K(x′, x i )include: Among them, γ is the adjustment factor, β j is the weight of the jth dimension, D is the dimension of the gesture feature vector, x′ j is the feature vector of the current gesture in the jth dimension, x i，j is the feature vector of the i-th gesture in the j-th dimension in the training dataset; The gesture label prediction model to which the gesture belongs is optimized, and the feature vector of the gesture during the helicopter lifeguard search and rescue operation is obtained in real time and input into the optimized gesture label prediction model to identify the gesture during the helicopter lifeguard search and rescue operation according to the operation category.

2. A method for recognizing hand gestures for helicopter lifeguard search and rescue operations according to claim 1, characterized in that: Optimizing the gesture label prediction model of the gesture includes: calculating the feature vector x of the i-th gesture in the training data set i Lagrange multiplier α on the c-th job category i，c and the bias term b for the cth job category c , specifically: Where C is the number of job categories, α j′，c is the feature vector x of the j′th gesture in the training dataset j′ Lagrange multiplier on the cth job category, y j′，c is the feature vector x of the j′th gesture in the training dataset j′ The gesture label corresponding to the c-th job category.

3. A method for recognizing hand gestures for helicopter lifeguard search and rescue operations according to claim 2, characterized in that: Set constraints, specifically: Among them, C′ is the regularization parameter used to control the calculation of α i，c and b c The complexity of time.

4. A helicopter lifeguard search and rescue operation gesture recognition system, characterized in that: include: A data acquisition module is used to acquire a training data set, wherein the training data set includes: a feature vector of a gesture and a gesture label, each of the gesture labels corresponds to a job category, wherein the feature vector of each gesture in the training data set corresponds to a job category of a gesture label; A training model module is used to set a gesture label prediction model to which a gesture belongs, obtain a feature vector of a current gesture, and obtain, based on the training data set, an operation category of a rescue operation to which the feature vector of the current gesture belongs; The gesture label prediction model to which the gesture belongs includes: Where f(x′) is the category of the rescue operation to which the feature vector x′ of the current gesture belongs, N is the number of samples in the training dataset, and α i，c is the feature vector x of the i-th gesture in the training dataset i Lagrange multiplier on the cth job category, y i，c is the feature vector x of the i-th gesture in the training dataset i The gesture label corresponding to the c-th job category is, if the feature vector x of the i-th gesture in the training dataset is i Belongs to the cth job category, then y i，c =1, otherwise 0, K(x′, x i ) is a kernel function used to calculate the feature vector x′ of the current gesture and the feature vector x of the i-th gesture in the training data set. i The similarity of b c is the bias term for the c-th job category; The kernel function K(x′, x i )include: Among them, γ is the adjustment factor, β j is the weight of the jth dimension, D is the dimension of the gesture feature vector, x′ j is the feature vector of the current gesture in the jth dimension, x i，j is the feature vector of the i-th gesture in the j-th dimension in the training dataset; The recognition module is used to optimize the gesture label prediction model to which the gesture belongs, and to obtain the feature vector of the gesture during the helicopter lifeguard search and rescue operation in real time, and input it into the optimized gesture label prediction model to identify the gesture during the helicopter lifeguard search and rescue operation according to the operation category.

5. A helicopter lifeguard search and rescue operation gesture recognition system as claimed in claim 4, characterized in that: Optimizing the gesture label prediction model of the gesture includes: calculating the feature vector x of the i-th gesture in the training data set i Lagrange multiplier α on the c-th job category i，c and the bias term b for the cth job category c , specifically: Where C is the number of job categories, α j′，c is the feature vector x of the j′th gesture in the training dataset j′ Lagrange multiplier on the cth job category, y j′，c is the feature vector x of the j′th gesture in the training dataset j′ The gesture label corresponding to the c-th job category.

6. A helicopter lifeguard search and rescue operation gesture recognition system as claimed in claim 5, characterized in that: Set constraints, specifically: Among them, C′ is the regularization parameter used to control the calculation of α i，c and b c The complexity of time.

Images