A non-contact game character control method for rehabilitation training
By capturing gesture features and establishing a mapping relationship between gestures and character movements using the Leap Motion device, the problem of the tedium of traditional hand rehabilitation training is solved, thereby improving the fun and effectiveness of rehabilitation training and providing an efficient human-computer interaction experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING UNIV OF TECH
- Filing Date
- 2026-03-10
- Publication Date
- 2026-07-24
AI Technical Summary
Traditional hand rehabilitation training methods are monotonous and boring, resulting in poor patient compliance and unsatisfactory rehabilitation outcomes.
A non-contact game character control method is adopted. Hand images are captured by the Leap Motion device, gesture features are extracted, and a mapping relationship between gestures and character actions is established. The gesture feature quantization strategy is used to control the game character's actions, achieving seamless conversion between gestures and character actions.
It significantly improves the effectiveness and enjoyment of rehabilitation training, increases patients' willingness to participate actively, and achieves an efficient and natural human-computer interaction experience.
Smart Images
Figure CN122455233A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of contactless game control technology, and in particular to a contactless game character control method for rehabilitation training. Background Technology
[0002] The incidence of hand diseases (such as tenosynovitis and carpal tunnel syndrome) is increasing year by year worldwide, seriously affecting patients' quality of life. Traditional hand rehabilitation training methods are monotonous and boring, resulting in poor patient compliance and unsatisfactory rehabilitation effects. Summary of the Invention
[0003] To enhance the effectiveness and enjoyment of rehabilitation training, this invention proposes a non-contact game character control method for rehabilitation training. The method includes: displaying the character movements of a preset puppet character during the game setup phase and prompting the user to make any gesture; continuously capturing multiple frames of hand images using a Leap Motion device and extracting gesture features, which are then associated and saved to a gesture feature library; creating a gesture recognition interface based on the gesture feature library and configuring a gesture feature quantization strategy to establish a mapping relationship between the gesture feature quantization strategy and the current character movement; and calling the gesture feature recognition interface during the game startup phase to perform gesture recognition and gesture feature quantization, and controlling the preset puppet character to perform corresponding actions based on the gesture feature quantization results and the mapping relationship.
[0004] In one or more embodiments, the game setting phase displays the character actions of a preset puppet character and prompts the user to make any gesture, including: sequentially displaying the preset puppet character's actions of jumping up and down to avoid obstacles, jumping up and down to grab airborne props, kicking a wine jar, and swaying left and right in the air; prompting the user to make any non-repeating gesture when each character action is displayed.
[0005] In one or more embodiments, the non-contact game character control method for rehabilitation training further includes: in response to the completion of the gesture recognition interface configuration for the currently displayed character action, displaying the next character action and performing corresponding gesture recognition interface creation and configuration; and in response to the completion of the gesture recognition interface creation and configuration for all character actions, starting the game.
[0006] In one or more embodiments, the Leap Motion device continuously captures multiple frames of hand images and extracts gesture features, which are then associated and saved to a gesture feature library. This includes: continuously capturing multiple frames of hand images and extracting finger joint features, wrist joint features, and / or palm features; determining dynamic features that have changed and static features that have not changed in the multiple frames of hand images; selecting all dynamic features and some static features, associating and saving them, and assigning gesture numbers to them for storage in the gesture feature library.
[0007] In one or more embodiments, creating a gesture recognition interface based on the gesture feature library and configuring a gesture feature quantization strategy to establish a mapping relationship between the gesture feature quantization strategy and the current character's action includes: creating a gesture recognition interface that binds to the latest saved associated features in the gesture feature library; configuring a complete quantization strategy or an amplitude quantization strategy based on the dynamic features in the associated features; configuring a stable quantization strategy based on the static features in the associated features; and obtaining a function call for the control command of the currently displayed character's action to establish a mapping relationship between the gesture feature quantization strategy and the current character's action.
[0008] In one or more embodiments, configuring a complete quantization strategy or an amplitude quantization strategy based on the dynamic features in the associated features includes: in response to the existence of multiple dynamic knuckle features in the associated features, calculating the minimum distance and the maximum distance between the multiple dynamic knuckle features; and constructing a normalization function based on the minimum distance and the maximum distance to evaluate the degree of completion.
[0009] In one or more embodiments, configuring a complete quantization strategy or an amplitude quantization strategy based on the dynamic features in the associated features further includes: in response to the presence of dynamic wrist joint features or palm features in the associated features, calculating the rotation angle of the wrist joint feature or the displacement range of the palm feature; and constructing a normalization function based on the rotation angle of the wrist joint feature or the displacement range of the palm feature to evaluate the amplitude.
[0010] In one or more embodiments, configuring a stability metric strategy based on static features among the associated features includes: in response to the existence of multiple static features among the associated features, calculating the rate of change of the relative distance between the multiple static features; and constructing a normalization function based on the rate of change of the relative distance between the static features to evaluate stability.
[0011] In one or more embodiments, during the game startup phase, the gesture feature recognition interface is invoked to perform gesture recognition and gesture feature quantization, and the preset puppet character is controlled to perform corresponding actions based on the gesture feature quantization result and the mapping relationship. This includes: after the game starts, the Leap Motion device continuously captures multiple frames of hand images and extracts finger joint features, wrist joint features, and / or palm features; the gesture feature recognition interface is sequentially invoked to perform feature matching on the extracted gesture features to recognize the gesture and confirm the matching gesture feature recognition interface; the collected gesture features are quantized according to the normalization function in the gesture feature recognition interface; the call function of the character action in the gesture feature recognition interface and the quantization result are encapsulated into control instructions and passed to the Unity animation engine through an internal message mechanism to control the preset puppet character to perform corresponding actions.
[0012] In one or more embodiments, the non-contact game character control method for rehabilitation training further includes: displaying special effects based on the execution results of the character's actions.
[0013] The beneficial effects of the present invention include: the present invention can automatically complete the arbitrary binding between gestures and character actions through the creation and configuration of the gesture feature recognition interface, and can automatically perform quantification processing according to the characteristics of gesture features and generate control commands with quantification results as parameters, thereby driving character actions in real time and efficiently, effectively improving the smoothness of gesture control, and thus significantly improving the rehabilitation training effect and user experience. Attached Figure Description
[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained based on these drawings without creative effort.
[0015] Figure 1 This is a flowchart illustrating the non-contact game character control method for rehabilitation training according to an embodiment of the present invention. Figure 2 This is a mapping diagram of the first gesture and the character's actions in an embodiment of the present invention; Figure 3 This is a mapping diagram of the second gesture and the character's actions in an embodiment of the present invention. Detailed Implementation
[0016] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to specific examples and the accompanying drawings.
[0017] To enhance the effectiveness and enjoyment of rehabilitation training, in one embodiment, this invention proposes a non-contact game character control method for rehabilitation training, such as... Figure 1 As shown, the method includes: Step S1: Displaying the character actions of a preset puppet character during the game setup phase and prompting the user to make any gesture; Step S2: Continuously capturing multiple frames of hand images using a Leap Motion device and extracting gesture features, which are then associated and saved to a gesture feature library; Step S3: Creating a gesture recognition interface based on the gesture feature library and configuring a gesture feature quantization strategy to establish a mapping relationship between the gesture feature quantization result and the current character action; Step S4: Calling the gesture feature recognition interface during the game startup phase to perform gesture recognition and gesture feature quantization, and controlling the preset puppet character to perform corresponding actions based on the gesture feature quantization result and the mapping relationship.
[0018] Specifically, the Leap Motion device can accurately capture the three-dimensional position, posture, gestures, and subtle movements of the user's hands, including fingertip position, knuckle bending angle, and palm orientation, using an infrared camera. This invention provides a hand rehabilitation training method for intangible cultural heritage puppetry. Its core technology lies in constructing a real-time gesture mapping method based on Leap Motion and Unity. Leap Motion accurately captures the skeletal joint data of the user's hand, and after feature acquisition and quantification, generates control commands that are input into a Unity-based game environment. This achieves a mapping between hand gestures and puppet character movements, enabling each rehabilitation gesture (such as grasping, pinching, and wrist movement) to trigger and play the corresponding puppet animation. This achieves a seamless transition from rehabilitation gestures to character movements, allowing hand rehabilitation training to be completed during gameplay, thus enhancing the effectiveness and enjoyment of rehabilitation training.
[0019] In one embodiment, to facilitate the development of different rehabilitation training movements for different users, the method of the present invention allows for the binding of arbitrary user gestures with character movements during the game setup phase, thereby adapting to the rehabilitation training needs of different users and greatly improving the practicality of the method. The specific implementation steps of the present invention are as follows: In one embodiment, the game setup phase displays the character actions of a preset puppet character and prompts the user to make any gesture, including: sequentially displaying the preset puppet character's actions of jumping up and down to avoid obstacles, jumping up and down to grab airborne props, kicking a wine jar, and swaying left and right in the air; prompting the user to make any non-repeating gesture when each character action is displayed.
[0020] Specifically, in this embodiment, when demonstrating a character's action, the operation command for that action is invoked and parameter binding is awaited. Optionally, parameter binding can be achieved by modifying the input parameters in the control command through macro definitions, or by modifying the interface bound to the calling function to receive different interface commands.
[0021] In one embodiment, based on the function of binding any gesture to a character's actions provided by the present invention, doctors can design rehabilitation gestures for users according to their actual rehabilitation needs and bind them to the corresponding character's actions.
[0022] In one embodiment, the method of the present invention further includes: in response to the completion of the gesture recognition interface configuration for the currently displayed character action, displaying the next character action and performing corresponding gesture recognition interface creation and configuration; and in response to the completion of gesture recognition interface creation and configuration for all character actions, starting the game. Optionally, the game's settings mode can be automatically entered upon initial power-on or upon user request, and the game will be launched directly on subsequent power-ons. Furthermore, the generation and configuration of the gesture feature recognition interface in this embodiment are both completed automatically; the specific process is described in the following embodiment.
[0023] In one embodiment, the Leap Motion device continuously captures multiple frames of hand images and extracts gesture features, which are then associated and saved to a gesture feature library. This includes: continuously capturing multiple frames of hand images and extracting finger joint features, wrist joint features, and / or palm features; determining dynamic features that have changed and static features that have not changed in the multiple frames of hand images; selecting all dynamic features and some static features, associating and saving them, and assigning gesture numbers to them for storage in the gesture feature library.
[0024] Specifically, the steps in this embodiment are executed synchronously with the character action display process, specifically in the background. During this process, the user makes specific gestures under the doctor's instruction, such as pinching with the index finger and thumb, clenching a fist, or waving the hands left and right. During each character action display, the user needs to repeatedly perform the same action until the gesture is successfully bound to the character action. Then, the next character action will be automatically displayed or the game will start. In the background, the LeapMotion device captures the skeletal node features of the user's gesture. For example, in the pinching gesture, the LeapMotion device captures the features of the finger joints, wrist joints, and / or palm, storing them chronologically. Then, by comparing the changes in the skeletal node features at different time points (e.g., changes in the distance between the index finger and thumb joints while the distances of other finger joints remain constant), the dynamic and static features can be distinguished. Finally, by extracting all dynamic features and some static features and associating them, the characteristics of the gesture are recorded and saved for subsequent recognition and association with character actions. The purpose of extracting only some static features is to reduce the number of feature matches in subsequent gesture steps, thereby improving the gesture recognition speed.
[0025] In one embodiment, creating a gesture recognition interface based on a gesture feature library and configuring a gesture feature quantization strategy to establish a mapping relationship between the gesture feature quantization result and the current character's action includes: creating a gesture recognition interface that binds to the latest saved associated features in the gesture feature library; configuring a complete quantization strategy or an amplitude quantization strategy based on the dynamic features in the associated features; configuring a stable quantization strategy based on the static features in the associated features; and obtaining the calling function of the control command for the currently displayed character's action to establish a mapping relationship between the gesture feature quantization strategy and the current character's action.
[0026] Specifically, the steps in this embodiment are executed synchronously with the character's action display process, specifically in the background. During this process, the newly created gesture recognition interface will be automatically bound to the latest generated associated features in the gesture feature library, and simultaneously obtain the calling function of the control command for the character's action. Thus, the mapping between gestures and character actions is achieved through the interface. The quantization strategy in this embodiment is used to compare the feature parameters of the gesture features collected during the game setup process with the feature parameters of the gesture features collected after the game starts to achieve quantization.
[0027] In one embodiment, during feature quantization, if both dynamic and static features exist, dynamic features are used first for quantization; however, during gesture recognition, both dynamic and static features are used for recognition, i.e., the complete associated features are used for recognition.
[0028] In one embodiment, configuring a complete quantization strategy or an amplitude quantization strategy based on dynamic features in the associated features includes: in response to the existence of multiple dynamic knuckle features in the associated features, calculating the minimum distance and the maximum distance between the multiple dynamic knuckle features; and constructing a normalization function based on the minimum distance and the maximum distance to evaluate the completion degree.
[0029] Specifically, in this embodiment, the normalization function constructed based on the minimum and maximum distances is as follows:
[0030] Among them, the distance between multiple dynamic knuckle features after the game starts is measured. The maximum distance between multiple dynamic knuckle features during the game setup phase. This represents the minimum distance between multiple dynamic knuckle features. Using the above normalization function, the distance between multiple dynamic knuckle features can be mapped to a value in the range [0,1]. Furthermore, when these multiple dynamic knuckle features originate from a pinching motion, the smaller the distance between the index finger and thumb, the larger the value of the normalization function.
[0031] In one embodiment, configuring a full quantization strategy or an amplitude quantization strategy based on dynamic features in the associated features further includes: in response to the presence of dynamic wrist joint features or palm features in the associated features, calculating the rotation angle of the wrist joint features or the displacement range of the palm features; and constructing a normalization function based on the rotation angle of the wrist joint features or the displacement range of the palm features to evaluate the amplitude.
[0032] Specifically, in this embodiment, a normalization function is constructed based on the rotation angle of the wrist joint features or the displacement range of the palm features as follows:
[0033] Among them, the rotation angle of the wrist joint features or the displacement range of the palm features after the game starts. The maximum rotation angle of the wrist joint feature or the maximum displacement range of the palm feature is set for the game stage. Through the above normalization function, the rotation angle of the wrist joint feature or the displacement range of the palm feature can be mapped to a value of [0,1]. When the rotation angle of the wrist joint feature or the displacement range of the palm feature comes from the hand swinging action, the larger the amplitude of the hand swinging action, the larger the value of the normalization function.
[0034] In one embodiment, configuring a stability metric strategy based on static features in the associated features includes: in response to the existence of multiple static features in the associated features, calculating the rate of change of the relative distance between the multiple static features; and constructing a normalization function based on the rate of change of the relative distance between the static features to evaluate stability.
[0035] Specifically, in this embodiment, a normalization function is constructed based on the rotation angle of the wrist joint features or the displacement range of the palm features as follows:
[0036] in, This represents the average distance between multiple static features during the game setup phase. This represents the distance between multiple static features during the game startup phase. N and n are the number of multiple static features during the game startup phase. The normalization function described above can map the rate of change of the distance between multiple static features to a value in the range of [0,1]. When the multiple static features are derived from a fist-clenching action, the closer the knuckles are to each other when clenching the fist, the larger the value of the normalization function.
[0037] In one embodiment, during the game startup phase, a gesture feature recognition interface is invoked to perform gesture recognition and gesture feature quantization, and a preset puppet character is controlled to perform corresponding actions based on the gesture feature quantization results and mapping relationships. This includes: after the game starts, the Leap Motion device continuously captures multiple frames of hand images and extracts finger joint features, wrist joint features, and / or palm features; the gesture feature recognition interface is sequentially invoked to perform feature matching on the extracted gesture features to recognize the gestures and confirm the matching gesture feature recognition interface; the collected gesture features are quantized according to the normalization function in the gesture feature recognition interface; the call function for the character's actions in the gesture feature recognition interface and the quantization results are encapsulated into control instructions and passed to the Unity animation engine through an internal message mechanism to control the preset puppet character to perform corresponding actions.
[0038] Specifically, in this embodiment, the mapping between gestures and character movements includes three key steps: first, specific hand gestures are identified using the LeapMotion gesture library; then, the recognition results are converted into corresponding animation control commands using the normalization and calling functions in the gesture feature recognition interface; and finally, the shadow puppet character is driven to perform preset performance actions. The system directly obtains gesture information through the gesture recognition interface provided by LeapMotion, eliminating the need for complex mathematical modeling and ensuring system stability and real-time performance. For complex rehabilitation movements such as immediately swinging after pinching or maintaining stability after grasping, the system identifies the changing trend of gesture features over time to confirm the intention of the movement, thereby achieving reliable detection of complex continuous movements.
[0039] In one embodiment, the method of the present invention further includes displaying special effects based on the execution result of the character's actions. Specifically, the execution result of the actions in this embodiment, such as the puppet character dodging obstacles, obtaining aerial props, or kicking and breaking a wine jar, will display encouraging special effects (such as flowing light or flower petals) when the above operations are successfully completed.
[0040] In one embodiment, the method of the present invention further includes: setting the level difficulty during the game setting stage to control the movement speed and movement range of obstacles, aerial props, or wine jars.
[0041] In some embodiments, the established mapping relationship between gestures and character actions includes: Example 1 The completion of the pinch gesture is calculated by normalizing the distance between the thumb and index finger according to the maximum opening distance and the minimum contact distance; when Leap Motion detects the "pinch" gesture, the system drives the fan in the virtual character's hand to perform an "opening and closing" animation.
[0042] Example 2 The amplitude of the left and right swaying gesture is evaluated based on the horizontal displacement range of the center of the palm; when the "left and right swaying" gesture is detected, the shadow puppet is controlled to perform the "kicking the wine jar" action.
[0043] Example 3 like Figure 2 and Figure 3 As shown, the stability of the grasping action is evaluated by statistically analyzing the feature change rate of the gesture in consecutive sampled frames. When the feature change is less than a preset threshold and maintains a certain continuity over time, the system considers the gesture recognition to be valid, that is, when the "grasping" gesture is confirmed, the shadow puppet character performs the action of "moving up and down to dodge the dart".
[0044] In some embodiments, the quantification results of the finger-pinching action are not only used to determine whether the "fan opening and closing" action is triggered, but also used as control parameters for the fan opening and closing angle, speed, and smoothness, realizing real-time mapping from the intensity of hand movements to the detailed expression of animation. The quantification results of the left and right swinging action are also used to dynamically adjust the force of the "kicking the wine jar" action, so that the animation effect can realistically reflect the changes in the force of the user's movements.
[0045] In one embodiment, the total latency from gesture recognition to virtual puppet motion generation is kept within tens of milliseconds, ensuring that the user's actual rehabilitation movements are highly consistent with the performance of the character on the screen, thereby achieving a stable, real-time, and natural human-computer interaction experience.
[0046] The technical effects of this invention include: 1. Establish an intuitive gesture-action mapping relationship to significantly enhance training motivation. By directly mapping basic rehabilitation gestures such as pinching, rocking, and grasping to characteristic role-playing actions such as opening and closing a fan, kicking a wine jar, and dodging darts, rehabilitation training is transformed from mechanical repetition into a meaningful performance experience, effectively stimulating patients' willingness to actively participate; 2. Achieve multi-dimensional real-time feedback based on movement quality. The system establishes a mapping mechanism between the quality of hand movements and puppet characters—the stability of the pinching gesture affects the smoothness of the fan opening and closing, the amplitude of the swaying motion determines the sense of power when kicking the wine jar, and the accuracy of the grasping motion relates to the success rate of dodging darts. The timely and intuitive visual feedback mechanism forces patients to concentrate on controlling the details of their movements and maintain a high level of focus during training.
[0047] 3. Optimize the human-computer interaction experience to ensure a smooth and natural training process. By directly integrating the LeapMotion gesture recognition interface, complex feature extraction and mathematical modeling processes are avoided, ensuring both high accuracy in gesture recognition and low-latency response in motion mapping. This natural and smooth interactive experience effectively reduces the user's learning cost and enhances the system's practicality and ease of use.
[0048] The above are exemplary embodiments disclosed in this invention. However, it should be noted that various changes and modifications can be made without departing from the scope of the embodiments of this invention as defined by the claims. The functions, steps, and / or actions of the methods according to the disclosed embodiments described herein do not need to be performed in any particular order. The embodiment numbers disclosed above are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.
[0049] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of the invention (including the claims) is limited to these examples. Within the framework of the invention, technical features of the above embodiments or different embodiments can be combined, and many other variations of different aspects of the invention exist, which are not provided in the details for the sake of brevity. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the invention should be included within the protection scope of the invention.
Claims
1. A non-contact game character control method for rehabilitation training, characterized in that, The method includes: During the game setup phase, display the character actions of the preset puppet characters and prompt the user to make any gesture; The Leap Motion device continuously captures multiple frames of hand images and extracts gesture features, which are then associated and saved to a gesture feature library. Create a gesture recognition interface based on the gesture feature library and configure a gesture feature quantization strategy to establish a mapping relationship between the gesture feature quantization strategy and the current character's actions; During the game startup phase, the gesture feature recognition interface is called to perform gesture recognition and gesture feature quantification, and the preset puppet character is controlled to perform corresponding actions based on the gesture feature quantification result and the mapping relationship.
2. The non-contact game character control method for rehabilitation training according to claim 1, characterized in that, During the game setup phase, the game displays the preset puppet character's actions and prompts the user to make any gesture, including: During the game setup phase, the preset puppet character's actions of jumping up and down to avoid obstacles, jumping up and down to grab airborne items, kicking a wine jar, and swaying left and right in the air are displayed in sequence. The system prompts users to make any gesture that is not repeated when each character's action is displayed.
3. The non-contact game character control method for rehabilitation training according to claim 2, characterized in that, The method further includes: Once the gesture recognition interface configuration for the currently displayed character's action is complete, the next character's action will be displayed, and the corresponding gesture recognition interface will be created and configured. The game starts in response to the creation and configuration of the gesture recognition interface that completes all character actions.
4. The non-contact game character control method for rehabilitation training according to claim 1, characterized in that, The Leap Motion device continuously captures multiple frames of hand images, extracts gesture features, and associates and saves them to a gesture feature library, including: The Leap Motion device continuously captures multiple frames of hand images and extracts finger joint features, wrist joint features, and / or palm features; Identify the dynamic features that have changed and the static features that have not changed in the multi-frame hand images; Select all dynamic features and some static features, associate and save them, assign gesture numbers, and store them in the gesture feature library.
5. The non-contact game character control method for rehabilitation training according to claim 4, characterized in that, Create a gesture recognition interface based on the gesture feature library and configure a gesture feature quantization strategy to establish a mapping relationship between the gesture feature quantization strategy and the current character's actions, including: Create a gesture recognition interface that binds to the latest saved associated features in the gesture feature library; Configure a complete quantization strategy or an amplitude quantization strategy based on the dynamic features in the aforementioned associated features; Configure a stable quantification strategy based on the static features in the associated features; The function calls the control commands for the currently displayed character's actions to establish a mapping relationship between the gesture feature quantization strategy and the current character's actions.
6. The non-contact game character control method for rehabilitation training according to claim 5, characterized in that, Configure a complete quantization strategy or magnitude quantization strategy based on the dynamic features in the associated features, including: In response to the presence of multiple dynamic knuckle features among the associated features, the minimum and maximum distances between the multiple dynamic knuckle features are calculated; A normalized function is constructed based on the minimum and maximum distances to evaluate the degree of completion.
7. The non-contact game character control method for rehabilitation training according to claim 5, characterized in that, Configuring a complete quantization strategy or an amplitude quantization strategy based on the dynamic features in the aforementioned associated features also includes: In response to the presence of a dynamic wrist joint feature or palm feature among the associated features, the rotation angle of the wrist joint feature or the displacement range of the palm feature is calculated. A normalized function is constructed based on the rotation angle of the wrist joint features or the displacement range of the palm features to evaluate the amplitude.
8. The non-contact game character control method for rehabilitation training according to claim 5, characterized in that, Configure a stable metric strategy based on the static features in the associated features, including: In response to the presence of multiple static features among the associated features, the rate of change of the relative distance between the multiple static features is calculated; A normalization function is constructed based on the rate of change of the relative distance between the static features to evaluate stability.
9. The non-contact game character control method for rehabilitation training according to any one of claims 5-8, characterized in that, During the game startup phase, the gesture feature recognition interface is invoked to perform gesture recognition and gesture feature quantization. Based on the gesture feature quantization results and the mapping relationship, the preset puppet character is controlled to perform corresponding actions, including: After the game starts, the Leap Motion device continuously captures multiple frames of hand images and extracts finger joint features, wrist joint features and / or palm features; The gesture feature recognition interface is called sequentially to perform feature matching on the extracted gesture features in order to recognize the gesture and confirm the matching gesture feature recognition interface; The collected gesture features are quantized according to the normalization function in the gesture feature recognition interface; The function call and quantization result of the character action in the gesture feature recognition interface are encapsulated into control instructions, which are then passed to the Unity animation engine through an internal message mechanism to control the preset puppet character to perform the corresponding actions.
10. The non-contact game character control method for rehabilitation training according to claim 9, characterized in that, The method further includes: Special effects are displayed based on the execution results of the character's actions.