A Dynamically Controllable Multi-Camera Data Acquisition System and Method

Abstract

This invention discloses a dynamic and controllable multi-camera data acquisition system and method, relating to the field of data acquisition technology. The system includes: establishing a three-dimensional model of a motion capture studio; capturing the movement trajectory of marked points on a motion capture suit during a performance by a motion capture actor, and obtaining the target point and the occlusion time period; acquiring feature points corresponding to the target points on the motion capture suit, and extracting target records from historical feature records; randomly selecting several time points from the target records to obtain the distance values ​​between the target point and the feature points at each time point, thereby obtaining the feature distance and its range; determining the warning value of the trajectory to be detected, and judging whether to prompt relevant technical personnel to repair the trajectory again. This invention, by analyzing the feature points corresponding to each target point and the corresponding historical feature records, determines whether the repair of the trajectory to be detected is reasonable and provides warning prompts, which helps to improve the accuracy and reliability of trajectory repair for marked points.

Description

Technical Field

[0001] This invention relates to the field of data acquisition technology, specifically a dynamic and controllable multi-camera data acquisition system and method. Background Technology

[0002] With the development of technology, motion capture, a key technology that records the movements of real people or objects and converts them into digital model motion data, is widely used in game development and animation creation to make virtual characters in games, animations, and films more lifelike. The main process of motion capture includes: actors wearing motion capture suits with marked points enter a motion capture studio, which contains multiple optical cameras and other monitoring equipment. These cameras capture the motion data of the actors at various marked points in real time during their performance. After capture, the raw data undergoes noise reduction and smoothing processes, and the processed data is then applied to the bound virtual character model to achieve highly realistic motion reproduction of the virtual character.

[0003] However, in actual shooting, the location of the marker point may be obscured, making it difficult to reconstruct the corresponding marker point position and resulting in missing image data. The existing solution is usually to simply predict the movement trajectory of the marker point at the obstruction point based on the position of the marker point before and after the obstruction. However, this method does not analyze historical records and connected marker points, which results in low accuracy of marker point restoration and unreliable restored movement trajectory. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic and controllable multi-camera data acquisition system and method to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A dynamic and controllable multi-camera data acquisition method includes the following steps:

[0007] Step S100: Establish a 3D model of the motion capture studio and mark the location of each monitoring device in the 3D model; based on the movement trajectory of the marked points on the motion capture suit captured by the monitoring devices when filming the performance of the motion capture actor wearing the motion capture suit, obtain the target point and the corresponding occlusion time period of the target point.

[0008] Step S200: Obtain the feature points corresponding to the target points on the motion capture suit, and obtain the historical feature records between the target points and the feature points. Based on the distance between the target points and the feature points in the 3D model, extract the target record from the feature records.

[0009] Step S300: Randomly select several time points from the time period corresponding to the target record, and obtain the distance value between the target point and the feature point at each time point. Based on the distance value, obtain the feature distance and feature distance range between the target point and the feature point.

[0010] Step S400: Repair the movement trajectory of the target point during the occluded period using the target repair method, and use the repaired trajectory as the trajectory to be detected; determine the warning value of the trajectory to be detected based on the feature distance and feature distance range between the target point and the feature point, and determine whether to prompt relevant technical personnel to repair it again based on the warning value.

[0011] Furthermore, step S100 includes: each monitoring device records the process of the motion capture actor wearing the motion capture suit and captures each marker point on the motion capture suit in real time; if at a certain time t, no more than U monitoring devices capture the position information of the marker point m, then the certain time t is taken as the occlusion time of the marker point m; if there are consecutive occlusion times for the marker point m, then the marker point m is taken as the target point, and the time period F corresponding to the consecutive occlusion times is taken as the occluded time period of the marker point m.

[0012] In actual motion capture, in order to realize the specific location of the marker point in three-dimensional space, it is generally necessary to include the marker point in two or more cameras. This is because a single camera can only acquire two-dimensional information of the marker point and cannot directly obtain the depth information of the marker point. However, when multiple cameras shoot the marker point from different angles, the specific location of the marker point in three-dimensional space can be calculated through the principle of triangulation.

[0013] Furthermore, step S200 includes:

[0014] Step S210: Obtain a target point q and its corresponding occlusion time period F. q The occlusion period F is obtained. q The earliest and latest times are determined by denoting the position of target point q at time T1 before the earliest time as P1. q The position of time T2 after the latest time is denoted as P2. q Obtain a marker point r on the motion capture suit. Randomly obtain several times when neither the target point q nor the marker point r is at their respective occlusion times. Obtain the distance between the target point q and the marker point r at each time. Calculate the variance based on the distance. If the variance is less than a preset first variance threshold, then the marker point r is taken as a feature point of the target point q, and the position of the marker point r at time T1 is recorded as P1. r The position at time T2 is denoted as P2. r ;

[0015] Step S220: Obtain the feature records corresponding to the target point q and feature point r during the historical movement process, take the total time period corresponding to the feature record as D, and satisfy that there is no occlusion time for the target point q and feature point r in the total time period D, and obtain the position corresponding to the target point q and feature point r at each time in the total time period D.

[0016] The period of time to be blocked F q The duration as D q Extract several segments of duration D from the total time period D. q For a given sub-feature record, the time period corresponding to that sub-feature record is designated as d, and the earliest time of time period d is designated as T1. d The latest time as T2 d And based on the target point q and the feature point r at time T1 d and time T2 d The location of [the location] is obtained at time T1. d and time T2 d At time q, the distance between target point q and feature point r is obtained; and the distance between target point q and feature point r at the earliest and latest times corresponding to each sub-feature record is obtained;

[0017] Step S230: Obtain the distance values ​​of the target point q and the feature point r at time T1, time T2, and the earliest and latest times corresponding to each sub-feature record, respectively. Calculate the variance corresponding to all distance values. If the variance is less than the preset second variance threshold, then the feature record is taken as the target record.

[0018] Further, step S300 includes: randomly obtaining the distance values ​​between the target point and the feature point at several times from the total time period corresponding to the target record, summing all the distance values ​​and calculating the average value as the feature distance C between the target point and the feature point, and obtaining the feature distance range between the target point and the feature point as: [K1×C, K2×C], where K1 and K2 are the first correlation coefficient and the second correlation coefficient, respectively, and 0 <K1<1<K2;

[0019] Furthermore, step S400 includes:

[0020] Step S410: Deploy inertial sensors at the target point location of the motion capture actor. Based on the inertial sensors, obtain the occlusion time F of a certain target point q. q The speed and direction of movement at each moment within the time frame will be occluded during the period F. qThe model is divided into S sub-time periods, each with a duration of D0, and ordered from 1 to S according to their chronological order. A spatial coordinate system corresponding to the 3D model is established. Based on the movement speed and direction at a certain moment within a sub-time period, the movement vector corresponding to that moment is obtained. This movement vector is then mapped onto the horizontal, vertical, and axial axes of the spatial coordinate system, and the corresponding coordinates are obtained. Based on the mapping results of the movement vector at each moment within a sub-time period onto the spatial coordinate system, the average coordinate values ​​on the horizontal, vertical, and axial axes are calculated. The vector obtained based on the average coordinate values ​​is used as the target vector for that sub-time period. Thus, the target vector for each sub-time period is obtained.

[0021] Step S420: Obtain the occluded time period F q The sub-time period F corresponding to a certain time a within the time period s If sub-time period F s The sequence number is not 1, based on the target point q during the occlusion period F. q The location P1 of the earliest time before time T1 q , will be at position P1 q Starting from position P2, move forward a distance of |V1|×D0 in the direction corresponding to the target vector V1 of sub-time period F1 (number 1), and set the position P2 accordingly, where |V1| is the magnitude of vector V1. Then, starting from position P2, move forward a distance of |V2|×D0 in the direction corresponding to the target vector V2 of sub-time period F2 (number 2), and set the position P3 accordingly. Continue this process to obtain sub-time periods F... s The previous sub-period F s-1 Corresponding position P s ;

[0022] Position P s Starting from the sub-time period F s Target vector V s Move forward in the corresponding direction |V s The position corresponding to the distance |×[a-D0(s-1)-T1] is taken as P a q And obtain the position P of the feature point r corresponding to the target point q at a certain time a. a r If position P a q and P a r The distance L between a Within the feature distance range, time a is taken as the feature time, and the position P corresponding to time a in the trajectory to be detected is obtained. a x Position P a q and P a xThe distance between them is taken as the target distance H at time a. a ;

[0023] Step S430: Then, N feature time points are obtained, and based on the target distance corresponding to each feature time point, a warning value for the trajectory to be detected is obtained. Where N is the number of characteristic moments, e is the natural constant, and H n The distance to the target at the nth feature time is given. If the warning value Y is greater than the preset warning threshold, relevant technical personnel are prompted to repair the trajectory of the target point q.

[0024] If a target point q is in the same position at both the earliest and latest times during the occluded period, the current target repair method might determine that the target point q is stationary during the occluded period. However, if the movement speed and direction of each sub-period of the occluded period are used to determine that the target point q is not stationary but is constantly moving, only to eventually return to the same position, then the trajectory to be detected can be judged to be abnormal. Therefore, the judgment method in this scheme, which first roughly determines the movement position of the marker point during the occluded period and then calculates the warning value for the trajectory to be detected, helps to improve the accuracy and reliability of the trajectory repair of the marker point.

[0025] A dynamic and controllable multi-camera data acquisition system includes a target point determination module, a target record extraction module, a feature distance calculation module, and an early warning module;

[0026] Target point determination module: used to build a 3D model of the motion capture studio, mark the location of each monitoring device in the 3D model; based on the movement trajectory of the marked points on the motion capture suit captured by the monitoring devices when the motion capture actors are performing, the target points and the corresponding occlusion time periods of the target points are obtained.

[0027] Target record extraction module: used to obtain the feature points corresponding to the target points on the motion capture suit, and obtain the historical feature records between the target points and the feature points. Based on the distance between the target points and the feature points in the 3D model, the target record is extracted from the feature records.

[0028] Feature distance calculation module: used to randomly select several moments from the time period corresponding to the target record, and obtain the distance value between the target point and the feature point at each moment. Based on the distance value, the feature distance and feature distance range between the target point and the feature point are obtained.

[0029] Early warning module: It is used to repair the movement trajectory of the target point during the period of occlusion by means of target repair method, and use the repaired trajectory as the trajectory to be detected; it determines the early warning value of the trajectory to be detected based on the feature distance and feature distance range between the target point and the feature point, and determines whether to prompt relevant technical personnel to repair it again based on the early warning value.

[0030] Furthermore, the target record extraction module includes a feature point determination unit, a distance value determination unit, and a target record extraction unit;

[0031] Feature point determination unit: used to acquire the target point and the corresponding occluded time period, obtain the location of the target point based on the earliest and latest times of the occluded time period, and determine the feature point from the marker point based on the location of the marker point;

[0032] Distance value determination unit: used to obtain the feature records corresponding to the target point and feature points during the historical movement process, extract several moments from the total time period corresponding to the feature records, and obtain the distance values ​​of the target point and feature points at each moment.

[0033] Target record extraction unit: It is used to calculate the variance corresponding to all distance values. If the variance is less than the preset second variance threshold, the feature record is taken as the target record.

[0034] Furthermore, the early warning module includes a target vector determination unit, a target distance determination unit, and an early warning unit;

[0035] Target vector determination unit: used to obtain the moving speed and direction of the target point at each moment during the occlusion period, and divide the occlusion period into S sub-periods of equal duration, and obtain the target vector corresponding to each sub-period based on the moving speed and direction;

[0036] Target distance determination unit: used to obtain the sub-time period corresponding to a certain moment within the occluded time period, and determine the characteristic time within the occluded time period based on the characteristic distance range, and then determine the target distance corresponding to each characteristic time.

[0037] Early warning unit: used to obtain the early warning value of the trajectory to be detected based on the target distance corresponding to each feature time; and to determine whether to prompt relevant technical personnel to repair it again based on the early warning value.

[0038] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention provides a dynamic and controllable multi-camera data acquisition system and method, including: establishing a three-dimensional model of a motion capture studio; capturing the movement trajectory of marked points on a motion capture suit when a motion capture actor performs, and obtaining the target point and the occluded time period; acquiring the feature points corresponding to the target points on the motion capture suit, and extracting the target record from historical feature records; randomly selecting several time points from the target records, obtaining the distance values ​​between the target point and the feature points at each time point, and then obtaining the feature distance and the feature distance range; determining the warning value of the trajectory to be detected, and judging whether to prompt relevant technical personnel to repair the trajectory again. This invention, by analyzing the feature points corresponding to each target point and the corresponding historical feature records, determines whether the repair of the trajectory to be detected is reasonable and provides warning prompts, which helps to improve the accuracy and reliability of the trajectory repair of marked points. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating a dynamic and controllable multi-camera data acquisition method according to the present invention.

[0040] Figure 2 This is a structural diagram of a dynamic and controllable multi-camera data acquisition system according to the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] Example: Figure 1 As shown, this invention provides a technical solution for a dynamically controllable multi-camera data acquisition method, comprising the following steps:

[0043] Step S100: Establish a 3D model of the motion capture studio and mark the location of each monitoring device in the 3D model; based on the movement trajectory of the marked points on the motion capture suit captured by the monitoring devices when filming the performance of the motion capture actor wearing the motion capture suit, obtain the target point and the corresponding occlusion time period of the target point.

[0044] Step S100 includes: each monitoring device records the performance process of the motion capture actor wearing the motion capture suit and captures each marker point on the motion capture suit in real time; if at a certain time t, no more than U monitoring devices capture the position information of a marker point m, then the time t is taken as the occlusion time of the marker point m; if there are consecutive occlusion times for a marker point m, then the marker point m is taken as the target point, and the time period F corresponding to the consecutive occlusion times is taken as the occluded time period of the marker point m. In this embodiment, U is 1.

[0045] Step S200: Obtain the feature points corresponding to the target points on the motion capture suit, and obtain the historical feature records between the target points and the feature points. Based on the distance between the target points and the feature points in the 3D model, extract the target record from the feature records.

[0046] Step S210: Obtain a target point q and its corresponding occlusion time period F. q The occlusion period F is obtained. q The earliest and latest times are determined by denoting the position of target point q at time T1 before the earliest time as P1. q The position of time T2 after the latest time is denoted as P2. q Obtain a marker point r on the motion capture suit. Randomly obtain several times when neither the target point q nor the marker point r is at their respective occlusion times. Obtain the distance between the target point q and the marker point r at each time. Calculate the variance based on the distance. If the variance is less than a preset first variance threshold, then the marker point r is taken as a feature point of the target point q, and the position of the marker point r at time T1 is recorded as P1. r The position at time T2 is denoted as P2. r ;

[0047] In this embodiment, taking the deployment of marker points at the wrist, elbow, and shoulder joints as an example, the distance between the wrist and elbow joints does not change much during actual movement due to skeletal structure, but the distance between the wrist and shoulder joints changes significantly. When the target point q is at the wrist and the marker point r is at the elbow joint, the elbow joint becomes a feature point of the wrist because the difference in distance is not significant. However, when the target point q is at the wrist and the marker point r is at the shoulder joint, the shoulder joint cannot become a feature point of the wrist because the difference in distance is too large. In this scheme, variance is used to characterize the magnitude of the distance difference.

[0048] Step S220: Obtain the feature records corresponding to the target point q and feature point r during the historical movement process, take the total time period corresponding to the feature record as D, and satisfy that there is no occlusion time for the target point q and feature point r in the total time period D, and obtain the position corresponding to the target point q and feature point r at each time in the total time period D.

[0049] The period of time to be blocked F q The duration as D q Extract several segments of duration D from the total time period D. q For a given sub-feature record, the time period corresponding to that sub-feature record is designated as d, and the earliest time of time period d is designated as T1. d The latest time as T2 d And based on the target point q and the feature point r at time T1 d and time T2 d The location of [the location] is obtained at time T1. d and time T2 d At time q, the distance between target point q and feature point r is obtained; and the distance between target point q and feature point r at the earliest and latest times corresponding to each sub-feature record is obtained;

[0050] Step S230: Obtain the distance values ​​of the target point q and the feature point r at time T1, time T2, and the earliest and latest times corresponding to each sub-feature record, respectively. Calculate the variance corresponding to all distance values. If the variance is less than the preset second variance threshold, then the feature record is taken as the target record.

[0051] In actual measurements, the distance between the wrist and elbow joint will normally have a slight error, but the error will not be too large. When the error is too large, it means that the recorded data is unreliable and cannot be used. Therefore, this step is to extract target records that can provide more accurate results for the following steps.

[0052] Step S300: Randomly select several time points from the time period corresponding to the target record, and obtain the distance value between the target point and the feature point at each time point. Based on the distance value, obtain the feature distance and feature distance range between the target point and the feature point.

[0053] Step S300 includes: randomly obtaining the distance values ​​between the target point and the feature point at several times from the total time period corresponding to the target record, summing all the distance values ​​and calculating the average value as the feature distance C between the target point and the feature point, and obtaining the feature distance range between the target point and the feature point as: [K1×C, K2×C], where K1 and K2 are the first correlation coefficient and the second correlation coefficient, respectively, and 0 <K1<1<K2;

[0054] Step S400: Repair the movement trajectory of the target point during the occluded period using the target repair method, and use the repaired trajectory as the trajectory to be detected; determine the warning value of the trajectory to be detected based on the feature distance and feature distance range between the target point and the feature point, and determine whether to prompt relevant technical personnel to repair it again based on the warning value.

[0055] Step S410: Deploy inertial sensors at the target point location of the motion capture actor. Based on the inertial sensors, obtain the occlusion time F of a certain target point q. q The speed and direction of movement at each moment within the time frame will be occluded during the period F. q The model is divided into S sub-time periods, each with a duration of D0, and ordered from 1 to S according to their chronological order. A spatial coordinate system corresponding to the 3D model is established. Based on the movement speed and direction at a certain moment within a sub-time period, the movement vector corresponding to that moment is obtained. This movement vector is then mapped onto the horizontal, vertical, and axial axes of the spatial coordinate system, and the corresponding coordinates are obtained. Based on the mapping results of the movement vector at each moment within a sub-time period onto the spatial coordinate system, the average coordinate values ​​on the horizontal, vertical, and axial axes are calculated. The vector obtained based on the average coordinate values ​​is used as the target vector for that sub-time period. Thus, the target vector for each sub-time period is obtained.

[0056] Step S420: Obtain the occluded time period F q The sub-time period F corresponding to a certain time a within the time period s If sub-time period F s The sequence number is not 1, based on the target point q during the occlusion period F. q The location P1 of the earliest time before time T1 q , will be at position P1 q Starting from position P2, move forward a distance of |V1|×D0 in the direction corresponding to the target vector V1 of sub-time period F1 (number 1), and set the position P2 accordingly, where |V1| is the magnitude of vector V1. Then, starting from position P2, move forward a distance of |V2|×D0 in the direction corresponding to the target vector V2 of sub-time period F2 (number 2), and set the position P3 accordingly. Continue this process to obtain sub-time periods F... s The previous sub-period F s-1 Corresponding position P s ;

[0057] Position P s Starting from the sub-time period F s Target vector V s Move forward in the corresponding direction |V s The position corresponding to the distance |×[a-D0(s-1)-T1] is taken as P a q And obtain the position P of the feature point r corresponding to the target point q at a certain time a. a r If position P a q and P a r The distance L between aWithin the feature distance range, time a is taken as the feature time, and the position P corresponding to time a in the trajectory to be detected is obtained. a x Position P a q and P a x The distance between them is taken as the target distance H at time a. a ;

[0058] Step S430: Then, N feature time points are obtained, and based on the target distance corresponding to each feature time point, a warning value for the trajectory to be detected is obtained. Where N is the number of characteristic moments, e is the natural constant, and H n The distance to the target at the nth feature time is given. If the warning value Y is greater than the preset warning threshold, relevant technical personnel are prompted to repair the trajectory of the target point q.

[0059] Formula y=1-e -x When x>0, y takes the value [0,1), and is a function of y increasing as x increases. Therefore, the range of the warning value Y in this scheme is 0 to 1. In this embodiment, when Y is greater than 0.6, it indicates that there is an anomaly in the trajectory to be detected, and relevant technicians need to repair the movement trajectory of the target point q.

[0060] This invention also provides a dynamically controllable multi-camera data acquisition system, such as... Figure 2 As shown, it includes: a target point determination module, a target record extraction module, a feature distance calculation module, and an early warning module;

[0061] Target point determination module: used to build a 3D model of the motion capture studio, mark the location of each monitoring device in the 3D model; based on the movement trajectory of the marked points on the motion capture suit captured by the monitoring devices when the motion capture actors are performing, the target points and the corresponding occlusion time periods of the target points are obtained.

[0062] Target record extraction module: used to obtain the feature points corresponding to the target points on the motion capture suit, and obtain the historical feature records between the target points and the feature points. Based on the distance between the target points and the feature points in the 3D model, the target record is extracted from the feature records.

[0063] Feature distance calculation module: used to randomly select several moments from the time period corresponding to the target record, and obtain the distance value between the target point and the feature point at each moment. Based on the distance value, the feature distance and feature distance range between the target point and the feature point are obtained.

[0064] Early warning module: It is used to repair the movement trajectory of the target point during the period of occlusion by means of target repair method, and use the repaired trajectory as the trajectory to be detected; it determines the early warning value of the trajectory to be detected based on the feature distance and feature distance range between the target point and the feature point, and determines whether to prompt relevant technical personnel to repair it again based on the early warning value.

[0065] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for acquiring data from multiple cameras based on dynamic and controllable systems, characterized in that, Includes the following steps: Step S100: Establish a three-dimensional model of the motion capture shed and mark the location of each monitoring device in the three-dimensional model; Based on the movement trajectory of the marked points on the motion capture suit captured by the monitoring equipment when the motion capture actor is performing, the target point and the corresponding time period of occlusion are obtained. Step S200: Obtain the feature points corresponding to the target points on the motion capture suit, and obtain the historical feature records between the target points and the feature points. Based on the distance between the target points and the feature points in the 3D model, extract the target record from the feature records. Step S300: Randomly select several time points from the time period corresponding to the target record, and obtain the distance value between the target point and the feature point at each time point. Based on the distance value, obtain the feature distance and feature distance range between the target point and the feature point. Step S400: Repair the movement trajectory of the target point during the occluded period using the target repair method, and use the repaired trajectory as the trajectory to be detected; Based on the characteristic distance and range between the target point and the feature point, the warning value of the trajectory to be detected is determined, and the warning value is used to determine whether to prompt relevant technical personnel to perform repairs again.

2. The method for acquiring data from multiple cameras based on dynamic controllability according to claim 1, characterized in that, Step S100 includes: each monitoring device records the process of the motion capture actor wearing the motion capture suit and captures each marker point on the motion capture suit in real time; if at a certain time t, no more than U monitoring devices capture the position information of the marker point m, then the certain time t is taken as the occlusion time of the marker point m; if there are consecutive occlusion times of the marker point m, then the marker point m is taken as the target point, and the time period F corresponding to the consecutive occlusion times is taken as the occluded time period of the marker point m.

3. The method for acquiring data from multiple cameras based on dynamic controllability according to claim 1, characterized in that, Step S200 includes: Step S210: Obtain a target point q and its corresponding occlusion time period F. q The occlusion period F is obtained. q The earliest and latest times are determined by denoting the position of target point q at time T1 before the earliest time as P1. q The position of time T2 after the latest time is denoted as P2. q A marker point r on the motion capture suit is obtained. Several times when neither the target point q nor the marker point r is at their respective occlusion times are randomly obtained. The distance between the target point q and the marker point r at each time is obtained, and the variance is calculated based on the distance. If the variance is less than a preset first variance threshold, the marker point r is taken as a feature point of the target point q, and the position of the marker point r at time T1 is recorded as P1. r The position at time T2 is denoted as P2. r ; Step S220: Obtain the feature records corresponding to the target point q and feature point r during the historical movement process, take the total time period corresponding to the feature record as D, and satisfy that there is no occlusion time for the target point q and feature point r in the total time period D, and obtain the position corresponding to the target point q and feature point r at each time in the total time period D. The period of time to be blocked F q The duration as D q Extract several segments of duration D from the total time period D. q For a given sub-feature record, the time period corresponding to that sub-feature record is designated as d, and the earliest time of time period d is designated as T1. d The latest time as T2 d And based on the target point q and the feature point r at time T1 d and time T2 d The location of [the location] is obtained at time T1. d and time T2 d At that time, the distance between the target point q and the feature point r is obtained; and the distance between the target point q and the feature point r at the earliest and latest times corresponding to each sub-feature record is obtained; Step S230: Obtain the distance values ​​of the target point q and the feature point r at time T1, time T2, and the earliest and latest times corresponding to each sub-feature record, respectively. Calculate the variance corresponding to all distance values. If the variance is less than a preset second variance threshold, then the feature record is taken as the target record.

4. The method for multi-camera data acquisition based on dynamic controllability according to claim 1, characterized in that, Step S300 includes: randomly obtaining the distance values ​​between the target point and the feature point at several times from the total time period corresponding to the target record, summing all the distance values ​​and calculating the average value as the feature distance C between the target point and the feature point, and obtaining the feature distance range between the target point and the feature point as: [K1×C, K2×C], where K1 and K2 are the first correlation coefficient and the second correlation coefficient, respectively, and 0 <K1<1<K2。 5. The method for multi-camera data acquisition based on dynamic controllability according to claim 1, characterized in that, Step S400 includes: Step S410: Deploy inertial sensors at the target point location of the motion capture actor. Based on the inertial sensors, obtain the occlusion time F of a certain target point q. q The speed and direction of movement at each moment within the time frame will be occluded during the period F. q The model is divided into S sub-time periods, each with a duration of D0, and ordered from 1 to S according to their chronological order. A spatial coordinate system corresponding to the 3D model is established. Based on the movement speed and direction at a certain moment within a sub-time period, the movement vector corresponding to that moment is obtained. The movement vector is then mapped onto the horizontal, vertical, and axial axes of the spatial coordinate system, and the corresponding coordinates are obtained. Based on the mapping results of the movement vector at each moment within a sub-time period onto the spatial coordinate system, the average coordinate values ​​on the horizontal, vertical, and axial axes are calculated. The vector obtained based on the average coordinate values ​​is used as the target vector corresponding to that sub-time period. Thus, the target vector corresponding to each sub-time period is obtained. Step S420: Obtain the occluded time period F q The sub-time period F corresponding to a certain time a within the time period s If sub-time period F s The sequence number is not 1, based on the target point q during the occlusion period F. q The location P1 of the earliest time before time T1 q , will be at position P1 q Starting from position P2, move forward a distance of |V1|×D0 in the direction corresponding to the target vector V1 of sub-time period F1 (number 1), and set the position P2 accordingly, where |V1| is the magnitude of vector V1. Then, starting from position P2, move forward a distance of |V2|×D0 in the direction corresponding to the target vector V2 of sub-time period F2 (number 2), and set the position P3 accordingly. Continue this process to obtain sub-time periods F... s The previous sub-period F s-1 Corresponding position P s ; Position P s Starting from the sub-time period F s Target vector V s Move forward in the corresponding direction |V s The position corresponding to the distance |×[a-D0(s-1)-T1] is taken as P a q And obtain the position P of the feature point r corresponding to the target point q at a certain time a. a r If position P a q and P a r The distance L between a Within the feature distance range, time a is taken as the feature time, and the position P corresponding to time a in the trajectory to be detected is obtained. a x Position P a q and P a x The distance between them is taken as the target distance H at time a. a ; Step S430: Then, N feature time points are obtained, and based on the target distance corresponding to each feature time point, a warning value for the trajectory to be detected is obtained. Where N is the number of characteristic moments, e is the natural constant, and H n The distance to the target at the nth feature time is given. If the warning value Y is greater than the preset warning threshold, relevant technical personnel are prompted to repair the trajectory of the target point q.

6. A multi-camera data acquisition system, used to execute the dynamically controllable multi-camera data acquisition method according to any one of claims 1-5, characterized in that, The system includes a target point determination module, a target record extraction module, a feature distance calculation module, and an early warning module; Target point determination module: used to build a 3D model of the motion capture studio, mark the location of each monitoring device in the 3D model; based on the movement trajectory of the marked points on the motion capture suit captured by the monitoring devices when the motion capture actors are performing, the target points and the corresponding occlusion time periods of the target points are obtained. Target record extraction module: used to obtain the feature points corresponding to the target points on the motion capture suit, and obtain the historical feature records between the target points and the feature points. Based on the distance between the target points and the feature points in the 3D model, the target record is extracted from the feature records. Feature distance calculation module: used to randomly select several moments from the time period corresponding to the target record, and obtain the distance value between the target point and the feature point at each moment. Based on the distance value, the feature distance and feature distance range between the target point and the feature point are obtained. Early warning module: It is used to repair the movement trajectory of the target point during the period of occlusion by means of target repair method, and use the repaired trajectory as the trajectory to be detected; it determines the early warning value of the trajectory to be detected based on the feature distance and feature distance range between the target point and the feature point, and determines whether to prompt relevant technical personnel to repair it again based on the early warning value.

7. A multi-camera data acquisition system according to claim 6, characterized in that, The target record extraction module includes a feature point determination unit, a distance value determination unit, and a target record extraction unit; Feature point determination unit: used to acquire the target point and the corresponding occluded time period, obtain the location of the target point based on the earliest and latest times of the occluded time period, and determine the feature point from the marker point based on the location of the marker point; Distance value determination unit: used to obtain the feature records corresponding to the target point and feature points during the historical movement process, extract several moments from the total time period corresponding to the feature records, and obtain the distance values ​​of the target point and feature points at each moment. Target record extraction unit: used to calculate the variance corresponding to all distance values. If the variance is less than a preset second variance threshold, the feature record is taken as the target record.

8. A multi-camera data acquisition system according to claim 6, characterized in that, The early warning module includes a target vector determination unit, a target distance determination unit, and an early warning unit. Target vector determination unit: used to obtain the moving speed and direction of the target point at each moment during the occlusion period, and divide the occlusion period into S sub-periods of equal duration, and obtain the target vector corresponding to each sub-period based on the moving speed and direction; Target distance determination unit: used to obtain the sub-time period corresponding to a certain moment within the occluded time period, and determine the characteristic time within the occluded time period based on the characteristic distance range, and then determine the target distance corresponding to each characteristic time. Early warning unit: used to obtain the early warning value of the trajectory to be detected based on the target distance corresponding to each feature time; and to determine whether to prompt relevant technical personnel to repair it again based on the early warning value.

Images