Animation character motion trajectory management system and method based on semantic analysis

Abstract

This invention discloses a semantic analysis-based animation character motion trajectory management system and method, belonging to the field of motion trajectory management technology. The system includes: acquiring the creator's animation character control commands; extracting semantic text from the creator's control commands; extracting action keywords; acquiring the creator's historical control records of the animation character; extracting target records from the control records to obtain a target template for the character to be controlled; acquiring historical animation video clips generated using the target template; extracting and analyzing the joint position movements of the animation character; intelligently extracting the animation keyframes corresponding to the target template; and automatically generating intermediate transition frames of the target video to form the animation character's motion trajectory. This invention analyzes control commands and the creator's historical control records, intelligently selecting target templates and animation keyframes, avoiding the inefficiency and unrealistic requirements of manual selection, and improving the reliability of animation character motion trajectory generation.

Description

Technical Field

[0001] This invention relates to the field of motion trajectory management technology, specifically to a motion trajectory management system and method for animated characters based on semantic analysis. Background Technology

[0002] Digital animation character motion trajectory generation is a core research direction in the modern digital entertainment and creative industries. It has irreplaceable application value in fields such as games, film and television, and virtual reality. User-driven character motion trajectory generation has become the mainstream development trend in the industry. However, the current technical system still has the following shortcomings: Existing solutions rely heavily on manual selection of preset motion trajectory templates and manual extraction of animation keyframes. This is not only inefficient, but also, because it does not take into account the characteristics of the animated character, the generated animation character motion trajectory often does not meet the actual creative needs. It cannot accurately match the creator's expression requirements and is difficult to meet the standards for the creation and application of high-quality digital content. Summary of the Invention

[0003] The purpose of this invention is to provide a semantic analysis-based animation character motion trajectory management system and method to solve the problems raised in the prior art.

[0004] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A semantic analysis-based method for managing the motion trajectory of animated characters includes the following steps: Obtain the creator's animation character control commands, extract the semantic text from the creator's control commands based on the input method, preprocess the semantic text, and extract action keywords; Obtain the creator's historical control records of animated characters, extract the character tags and application scene tags of the animated characters corresponding to the control records, extract the target records from the control records based on the action keywords and the application scene of the current character to be controlled, and use the motion trajectory template that is used most frequently in the target records as the target template of the current character to be controlled. Acquire historical animation video clips generated using the target template, build 3D models of the corresponding animation characters for the video clips, extract and analyze the joint position movement of the animation characters, and intelligently extract the animation keyframes corresponding to the target template; Animation keyframes are frames in an animated video that represent key postures and changes in motion of a character. They are usually frames where the character undergoes key posture changes or abrupt changes in motion, such as a sudden increase in speed or a sudden turn. Using animation keyframes helps to simplify the amount of animation data, reduce subsequent computational pressure, and at the same time retain the core features and abrupt changes of the character's motion, thereby improving the efficiency of animation generation and the smoothness of the trajectory. In this scheme, joint position changes are used to obtain animation keyframes, which can better capture the subtle changes and abrupt changes in the character's motion, accurately reflect posture changes, and improve the accuracy and representativeness of keyframe extraction. An animated video of the application scene where the character to be controlled is generated using the target template is used as the target video; interpolation calculations are performed based on the key frames of the animation to automatically generate intermediate transition frames of the target video, forming the motion trajectory of the animated character.

[0005] Furthermore, extract action keywords, including: Obtain the creator's control commands for the animated character; the input methods for these control commands include voice input and text input. Pre-deploy surveillance cameras on the computer; During voice input, there may be scenarios where the creator communicates with others or makes calls away from the scene. In these cases, the voice issued by the creator is not a control command for the animated character, and therefore should not be recognized as a valid control command. Therefore, this solution addresses the above problem by first filtering and extracting the voice that belongs to the valid control command, and then performing semantic text extraction on the valid voice. The specific analysis process is as follows: Before inputting control commands, the creator needs to verify their identity. After verification, they enter the animation character control system. If the control command input method is voice input, the monitoring segment during voice input is captured, and the creator's face area is captured. The number of ROI regions on the face area during voice input is obtained. Any one of the creator's lip area, throat area, jaw area, and nasal wing area is taken as the target area. The edge contour changes of the target area are analyzed, and the degree of change of the target area is calculated. If the number of ROI regions exceeds a preset threshold and the degree of change exceeds a preset threshold, the input speech is converted into semantic text, and action keywords are extracted from the semantic text.

[0006] Furthermore, calculating the degree of change in the target area includes: acquiring any two adjacent frames from the monitoring segment and treating them as a frame set; detecting the outer edge contours of the target area in the preceding and following frames respectively; performing normalization and alignment operations on the two outer edge contours; and obtaining the maximum area M of the region enclosed by the two outer edge contours. maxThe target level of the image set is obtained by considering the intersection area M of the regions enclosed by the two outer edge contours. Calculate the target intensity of several image sets and take the average value to obtain the degree of change in the target area.

[0007] The normalization alignment operation of the edge contour includes: calculating the geometric center, minimum bounding rectangle and principal axis direction of the contour respectively; achieving position alignment through translation transformation; achieving size normalization through scaling; and achieving angle alignment through rotation transformation, so that the two edge contours are in the same position, scale and angle. Since the normalization alignment operation of the edge contour is existing technology, it will not be described in detail here.

[0008] Furthermore, the target record is extracted from the control log, including: Get all role tags and all scene tags corresponding to control record R, and establish a first role tag set and a first scene tag set respectively; get all role tags of the currently controlled role when it was created, and all scene tags of the application scene in which the controlled role is currently located, and establish a second role tag set and a second scene tag set respectively. The first role tag set and the second role tag set have the same number of elements, both of which are NC; the first scene tag set and the second scene tag set have the same number of elements, both of which are NS. Preset the animation character weight W1 and application scene weight W2, collect the number of identical labels NC0 in the first character label set and the second character label set, and the number of identical labels NS0 in the first scene label set and the second scene label set, and obtain the reliability of the control record R as: If the reliability is greater than the preset reliability threshold, and the actions of the animated characters in the control record R match the extracted action keywords, then the control record R is used as the target record to obtain all target records.

[0009] Furthermore, the system intelligently extracts the animation keyframes corresponding to the target template, including: Retrieve historical animation video clips generated using the target template, obtain the 3D model of the corresponding animation character, and extract the location of all joints of the animation character; The animated video clip is divided into a set of frames for each consecutive H0 frame, resulting in several sets of frames. Based on the changes in the joint positions within the set of frames, the target frame is obtained, and the target frame is used as the animation keyframe corresponding to the target template.

[0010] Further, obtaining the target frame in the frame set includes: acquiring all frames in a certain frame set G; determining the first movement vector of each joint based on the earliest two frames, where the first movement vector is the vector from the position of the joint in the previous frame to the position of the joint in the next frame; determining the second movement vector of each joint based on the last two frames, where the second movement vector is the vector from the position of the joint in the previous frame to the position of the joint in the next frame; if there is at least one joint whose difference between the magnitudes of its first and second movement vectors is greater than a preset magnitude threshold, or whose angle between the directions of its first and second movement vectors is greater than a preset angle threshold, marking the frame set G to obtain all marked frame sets; and taking the middle frame of each marked frame set as the target frame.

[0011] The semantic analysis-based animation character motion trajectory management system includes a motion keyword extraction module, a target template determination module, an animation keyframe extraction module, and a motion trajectory formation module. Action Keyword Extraction Module: Used to obtain the creator's animation character control commands, extract semantic text from the creator's control commands based on the input method, preprocess the semantic text, and extract action keywords; Target template determination module: used to obtain the creator's historical control records of animation characters, extract the character tags and application scene tags of the animation characters corresponding to the control records, extract the target records from the control records based on the action keywords and the application scene of the current character to be controlled, and use the motion trajectory template that is used most frequently in the target records as the target template of the current character to be controlled. Animation Keyframe Extraction Module: Used to acquire animation video clips generated using the target template in the past, build a 3D model of the animation character corresponding to the animation video clip, extract and analyze the joint position movement of the animation character, and intelligently extract the animation keyframes corresponding to the target template; Motion trajectory formation module: This module uses the animated video of the application scene where the character to be controlled is located, generated using the target template, as the target video; it performs interpolation calculations based on the animation keyframes to automatically generate intermediate transition frames of the target video, thus forming the motion trajectory of the animated character.

[0012] Furthermore, the target template determination module includes a tag set establishment unit and a target template determination unit; Tag set establishment unit: used to obtain all role tags and all scene tags corresponding to the control record, and establish the first role tag set and the first scene tag set respectively; obtain all role tags of the currently controlled role when it was created, and all scene tags of the application scene in which the controlled role is currently located, and establish the second role tag set and the second scene tag set respectively. Target template determination unit: used to preset the weights of animated characters and application scenarios, and obtain the reliability of control records based on the first character tag set, the second character tag set, the first scene tag set, and the second scene tag set, thereby determining the target record.

[0013] Furthermore, the animation keyframe extraction module includes an animation keyframe extraction unit; Animation keyframe extraction unit: used to acquire historical animation video clips generated using the target template, acquire the 3D model of the corresponding animation character, and extract the location of all joints of the animation character; The animated video clip is divided into a set of frames for each consecutive H0 frame, resulting in several sets of frames. Based on the changes in the joint positions within the set of frames, the target frame is obtained, and the target frame is used as the animation keyframe corresponding to the target template.

[0014] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention provides an animation character motion trajectory management system and method based on semantic analysis, including: acquiring the creator's animation character control commands, extracting semantic text from the creator's control commands, and extracting action keywords; acquiring the creator's historical control records of the animation character, extracting target records from the control records, and obtaining a target template for the character to be controlled; acquiring historical animation video clips generated using the target template, extracting and analyzing the joint position movements of the animation character, and intelligently extracting the animation keyframes corresponding to the target template; and automatically generating intermediate transition frames of the target video to form the animation character's motion trajectory. This invention, by analyzing control commands and the creator's historical control records, intelligently selects target templates and animation keyframes, avoiding the inefficiency and incompatibility with actual creative needs caused by manual selection, and improving the reliability of animation character motion trajectory generation. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages 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 drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating the semantic analysis-based animation character motion trajectory management method of the present invention. Detailed Implementation

[0017] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims.

[0018] Example: Figure 1 As shown, this invention provides a technical solution for managing the motion trajectory of animated characters based on semantic analysis, including the following steps: The system obtains the creator's control commands for the animated character. These commands can be input via voice or text. In this embodiment, voice input is performed by the creator using a microphone, and text input is performed by typing. However, during voice input, there may be scenarios where the creator is communicating with others or making calls away from the scene. In these cases, the voice output by the creator is not a control command for the animated character and should not be recognized as a valid control command. Therefore, this solution addresses this issue by first filtering and extracting voice commands that are valid control commands, and then performing semantic text extraction on the valid voice commands. The specific analysis process is as follows: Pre-deploy surveillance cameras on the computer; Creators need to authenticate themselves before inputting control commands, and after authentication, they enter the animation character control system.

[0019] If the control command input method is voice input, a monitoring segment during voice input is captured, and the creator's face region is captured. The number of ROI regions on the face region during voice input is obtained. In this embodiment, the ROI region is the region of interest, specifically the area where facial organs are located, including the eyes, ears, and mouth. Any one of the creator's lip region, throat region, jaw region, and nasal region is used as the target region. The target region is the area that moves when the creator speaks, used to determine whether the creator is speaking. When speaking, the edge contour of the target region changes. The changes in the edge contour of the target region are analyzed, and the degree of change in the target region is calculated, including: Acquire any two adjacent frames from the surveillance footage and treat them as a single frame set. Detect the outer edge contours of the target region in both the preceding and following frames. Perform normalization and alignment operations on the two outer edge contours and obtain the maximum area M within the region enclosed by the two outer edge contours. max The target level of the image set is obtained by considering the intersection area M of the regions enclosed by the two outer edge contours. The greater the degree of target intensity, the more it indicates that the target area is changing, proving that the creator is speaking; calculate the degree of target intensity of several sets of images and take the average value to obtain the degree of change of the target area.

[0020] The normalization alignment operation of the edge contour includes: calculating the geometric center, minimum bounding rectangle and principal axis direction of the contour respectively; achieving position alignment through translation transformation; achieving size normalization through scaling; and achieving angle alignment through rotation transformation, so that the two edge contours are in the same position, scale and angle. Since the normalization alignment operation of the edge contour is existing technology, it will not be described in detail here.

[0021] If the number of ROI regions exceeds a preset threshold and the degree of change exceeds a preset degree threshold, the input speech is converted into semantic text. The more ROI regions captured, the more likely the creator is facing the computer rather than leaving or communicating with others. The greater the degree of change, the more likely the creator is speaking. Therefore, when the number of ROI regions exceeds the preset threshold and the degree of change exceeds the preset degree threshold, it proves that the creator is inputting control commands. At this time, the input speech can be converted into semantic text based on existing speech recognition technology.

[0022] The semantic text is preprocessed, including text cleaning, stop word removal, and text standardization. The preprocessing process is existing technology and will not be described in detail here. Action keywords are extracted from the semantic text. In this embodiment, motion keywords include forward, backward, left, and jump.

[0023] Retrieve the creator's historical control records of animated characters, extract the character tags and application scene tags corresponding to the control records, including character tags such as body type, style, and attributes, and scene tags such as scene type, environment style, weather conditions, and lighting conditions. Based on action keywords and the application scene of the currently controlled character, extract the target record from the control records, and use the most frequently used motion trajectory template in the target record as the target template for the currently controlled character. Specifically: Get all role tags and all scene tags corresponding to control record R, and establish a first role tag set and a first scene tag set respectively; get all role tags of the currently controlled role when it was created, and all scene tags of the application scene in which the controlled role is currently located, and establish a second role tag set and a second scene tag set respectively. The first role tag set and the second role tag set have the same number of elements, both of which are NC; the first scene tag set and the second scene tag set have the same number of elements, both of which are NS. Preset the animation character weight W1 and application scene weight W2, collect the number of identical labels NC0 in the first character label set and the second character label set, and the number of identical labels NS0 in the first scene label set and the second scene label set, and obtain the reliability of the control record R as: If the reliability is greater than the preset reliability threshold, and the actions of the animated characters in the control record R match the extracted action keywords, then the control record R is used as the target record to obtain all target records.

[0024] In this scheme, NC0 / NC represents the similarity between the tags in the first and second character tag sets. The larger the NC0 / NC, the more similar the animated character in the control record R is to the character to be controlled. NS0 / NS represents the similarity between the tags in the first and second scene tag sets. The larger the NS0 / NS, the more similar the application scene in the control record R is to the application scene of the character to be controlled. In this scheme, both W1 and W2 are greater than 0, indicating higher reliability. This means that the control record R matches the character to be controlled by the current creator and the current application scene more closely, and should be used as the target record. If multiple target records use the same motion trajectory template, and it is used the most, it means that the creator is more inclined to use this motion trajectory template at present. Therefore, this motion trajectory template can be used as the target template, reducing the complexity of manual operation and improving creative efficiency.

[0025] The process involves acquiring historical animation video clips generated using the target template, creating 3D models of the corresponding animated characters, extracting and analyzing the joint position movements of the animated characters, and intelligently extracting the animation keyframes corresponding to the target template. Specifically: Retrieve historical animation video clips generated using the target template, obtain the 3D model of the corresponding animation character, and extract the location of all joints of the animation character; The animated video clip is divided into several frame sets, each consisting of consecutive H0 frames. H0 is an odd number greater than or equal to 3. To ensure more accurate keyframe extraction, H0 should not be set too large; in this embodiment, H0 = 3 or 5. Based on the joint position changes within the frame sets, the target frames are obtained. These target frames are then used as the animation keyframes corresponding to the target template. Specifically: Acquire all frames in a certain frame set G. Based on the earliest two frames, determine the first movement vector of each joint. The first movement vector is the vector from the position of the joint in the previous frame to the position in the next frame. Based on the last two frames, determine the second movement vector of each joint. The second movement vector is the vector from the position of the joint in the previous frame to the position in the next frame. If there is at least one joint whose difference between the magnitudes of its first and second movement vectors is greater than a preset magnitude threshold, or whose angle between the directions of its first and second movement vectors is greater than a preset angle threshold, mark the frame set G to obtain all marked frame sets. Take the center frame of each marked frame set as the target frame. For H0=3, the center frame refers to the second frame. For H0=5, the center frame refers to the third frame.

[0026] An animated video of the application scene where the character to be controlled is located, generated using a target template, will be used as the target video. Interpolation calculations will be performed based on the animation keyframes to automatically generate intermediate transition frames in the target video, forming the motion trajectory of the animated character. Interpolation calculation methods include existing algorithms such as linear interpolation, Bézier curves, elastic easing, and SLERP interpolation. The appropriate algorithm should be selected based on the actual situation. All of the above algorithms are existing technologies and will not be elaborated here.

[0027] This invention analyzes control commands and the creator's historical control records to intelligently select target templates and animation keyframes, avoiding the inefficiency and incompatibility with actual creative needs caused by manual selection, and helps to improve the reliability of animation character motion trajectory generation.

[0028] This embodiment also provides a semantic analysis-based animation character motion trajectory management system, including an action keyword extraction module, a target template determination module, an animation keyframe extraction module, and a motion trajectory formation module; Action Keyword Extraction Module: Used to obtain the creator's animation character control commands, extract semantic text from the creator's control commands based on the input method, preprocess the semantic text, and extract action keywords; Target template determination module: used to obtain the creator's historical control records of animation characters, extract the character tags and application scene tags of the animation characters corresponding to the control records, extract the target records from the control records based on the action keywords and the application scene of the current character to be controlled, and use the motion trajectory template that is used most frequently in the target records as the target template of the current character to be controlled. Animation Keyframe Extraction Module: Used to acquire animation video clips generated using the target template in the past, build a 3D model of the animation character corresponding to the animation video clip, extract and analyze the joint position movement of the animation character, and intelligently extract the animation keyframes corresponding to the target template; Motion trajectory formation module: This module uses the animated video of the application scene where the character to be controlled is located, generated using the target template, as the target video; it performs interpolation calculations based on the animation keyframes to automatically generate intermediate transition frames of the target video, thus forming the motion trajectory of the animated character.

[0029] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0030] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0031] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for managing the motion trajectory of animated characters based on semantic analysis, characterized in that, Includes the following steps: Obtain the creator's animation character control commands, extract the semantic text from the creator's control commands based on the input method, preprocess the semantic text, and extract action keywords; Obtain the creator's historical control records of animated characters, extract the character tags and application scene tags of the animated characters corresponding to the control records, extract the target records from the control records based on the action keywords and the application scene of the current character to be controlled, and use the motion trajectory template that is used most frequently in the target records as the target template of the current character to be controlled. Acquire historical animation video clips generated using the target template, build 3D models of the corresponding animation characters for the video clips, extract and analyze the joint position movement of the animation characters, and intelligently extract the animation keyframes corresponding to the target template; The target video will be an animated video of the application scene in which the character to be controlled is located, generated using the target template. Interpolation calculations are performed based on animation keyframes to automatically generate intermediate transition frames in the target video, forming the motion trajectory of the animated character.

2. The method for managing the motion trajectory of animated characters based on semantic analysis according to claim 1, characterized in that, Extract action keywords, including: Obtain the creator's control commands for the animated character; the input methods for these control commands include voice input and text input. Pre-deploy surveillance cameras on the computer; Before inputting control commands, the creator needs to verify their identity. After verification, they enter the animation character control system. If the control command input method is voice input, the monitoring segment during voice input is captured, and the creator's face area is captured. The number of ROI regions on the face area during voice input is obtained. Any one of the creator's lip area, throat area, jaw area, and nasal wing area is taken as the target area. The edge contour changes of the target area are analyzed, and the degree of change of the target area is calculated. If the number of ROI regions exceeds a preset threshold and the degree of change exceeds a preset threshold, the input speech is converted into semantic text, and action keywords are extracted from the semantic text.

3. The method for managing the motion trajectory of animated characters based on semantic analysis according to claim 2, characterized in that, Calculating the degree of change in the target area includes: acquiring any two adjacent frames from the monitoring segment and treating them as a frame set; detecting the outer edge contours of the target area in the preceding and following frames respectively; performing normalization and alignment operations on the two outer edge contours; and obtaining the maximum area M of the region enclosed by the two outer edge contours. max The target level of the image set is obtained by considering the intersection area M of the regions enclosed by the two outer edge contours. Calculate the target intensity of several image sets and take the average value to obtain the degree of change in the target area.

4. The method for managing the motion trajectory of animated characters based on semantic analysis according to claim 1, characterized in that, Extract the target record from the control log, including: Get all role tags and all scene tags corresponding to control record R, and establish a first role tag set and a first scene tag set respectively; get all role tags of the currently controlled role when it was created, and all scene tags of the application scene in which the controlled role is currently located, and establish a second role tag set and a second scene tag set respectively. The first role tag set and the second role tag set have the same number of elements, both of which are NC; the first scene tag set and the second scene tag set have the same number of elements, both of which are NS. Preset the animation character weight W1 and application scene weight W2, collect the number of identical labels NC0 in the first character label set and the second character label set, and the number of identical labels NS0 in the first scene label set and the second scene label set, and obtain the reliability of the control record R as: If the reliability is greater than the preset reliability threshold, and the actions of the animated characters in the control record R match the extracted action keywords, then the control record R is used as the target record to obtain all target records.

5. The method for managing the motion trajectory of animated characters based on semantic analysis according to claim 1, characterized in that, Intelligent extraction of animation keyframes corresponding to the target template, including: Retrieve historical animation video clips generated using the target template, obtain the 3D model of the corresponding animation character, and extract the location of all joints of the animation character; The animated video clip is divided into a set of frames for each consecutive H0 frame, resulting in several sets of frames. Based on the changes in the joint positions within the set of frames, the target frame is obtained, and the target frame is used as the animation keyframe corresponding to the target template.

6. The method for managing the motion trajectory of animated characters based on semantic analysis according to claim 5, characterized in that, Obtaining the target frame from a set of frames includes: acquiring all frames in a certain set of frames G; determining the first movement vector of each joint based on the earliest two frames, wherein the first movement vector is a vector pointing from the position of the joint in the previous frame to the position of the joint in the next frame; determining the second movement vector of each joint based on the last two frames, wherein the second movement vector is a vector pointing from the position of the joint in the previous frame to the position of the joint in the next frame; if there is at least one joint whose difference between the magnitudes of its first movement vector and its second movement vector is greater than a preset magnitude threshold, or whose angle between the directions of its first movement vector and its second movement vector is greater than a preset angle threshold, marking the set of frames G to obtain all marked frame sets; and taking the middle frame of each marked frame set as the target frame.

7. A semantic analysis-based animation character motion trajectory management system, used to execute the semantic analysis-based animation character motion trajectory management method according to any one of claims 1-6, characterized in that, The system includes an action keyword extraction module, a target template determination module, an animation keyframe extraction module, and a motion trajectory formation module; Action Keyword Extraction Module: Used to obtain the creator's animation character control commands, extract semantic text from the creator's control commands based on the input method, preprocess the semantic text, and extract action keywords; Target template determination module: used to obtain the creator's historical control records of animation characters, extract the character tags and application scene tags of the animation characters corresponding to the control records, extract the target records from the control records based on the action keywords and the application scene of the current character to be controlled, and use the motion trajectory template that is used most frequently in the target records as the target template of the current character to be controlled. Animation Keyframe Extraction Module: Used to acquire animation video clips generated using the target template in the past, build a 3D model of the animation character corresponding to the animation video clip, extract and analyze the joint position movement of the animation character, and intelligently extract the animation keyframes corresponding to the target template; Motion trajectory formation module: Used to take the animated video of the application scene where the character to be controlled is located, generated using the target template, as the target video; Interpolation calculations are performed based on animation keyframes to automatically generate intermediate transition frames in the target video, forming the motion trajectory of the animated character.

8. A multimodal physiological data monitoring system according to claim 7, characterized in that, The target template determination module includes a tag set establishment unit and a target template determination unit; Tag set establishment unit: used to obtain all role tags and all scene tags corresponding to the control record, and establish the first role tag set and the first scene tag set respectively; obtain all role tags of the currently controlled role when it was created, and all scene tags of the application scene in which the controlled role is currently located, and establish the second role tag set and the second scene tag set respectively. Target template determination unit: used to preset the weights of animated characters and application scenarios, and obtain the reliability of control records based on the first character tag set, the second character tag set, the first scene tag set, and the second scene tag set, thereby determining the target record.

9. A multimodal physiological data monitoring system according to claim 7, characterized in that, The animation keyframe extraction module includes an animation keyframe extraction unit; Animation keyframe extraction unit: used to acquire historical animation video clips generated using the target template, acquire the 3D model of the corresponding animation character, and extract the location of all joints of the animation character; The animated video clip is divided into a set of frames for each consecutive H0 frame, resulting in several sets of frames. Based on the changes in the joint positions within the set of frames, the target frame is obtained, and the target frame is used as the animation keyframe corresponding to the target template.

Images