Three-dimensional animation motion data resource management and generation system

Abstract

A three-dimensional animation motion data resource management and generation system belongs to the technical field of three-dimensional animation motion data, and comprises the following steps: executing unreasonable detection on three-dimensional animation motion data by using a Z-score method to obtain unreasonable data, and constructing an unreasonable data set; calculating clutter value representation of the unreasonable data in the unreasonable data group and clutter value representation of the unreasonable data in a near area to calibrate Z scores of the unreasonable data in the Z-score method, and obtaining characteristic Z scores of the unreasonable data; if the characteristic Z score is higher than the score critical quantity, determining that the target unreasonable data is correct unreasonable data, and if the characteristic Z score is lower than the score critical quantity, determining that the target unreasonable data is a clutter value; finally confirming that the rocker is in a failure state; therefore, the restricted Boltzmann machine used for advanced estimation is used for executing advanced estimation of the unreasonable data of the time point queue, and the unreasonable data estimated in advance are accurately obtained effectively through queue information and unreasonable mark information.

Description

Technical Field

[0001] The invention belongs to the technical field of three-dimensional animation motion data, and in particular relates to a three-dimensional animation motion data resource management and generation system. Background Art

[0002] The three-dimensional animation motion data is just like the three-dimensional motion data of the joystick sensed by the motion sensor in the TV game or computer game in the prior art solution with patent publication number "CN101721812B", that is, the three-dimensional motion data of the joystick is the three-dimensional animation motion data.

[0003] Specifically, the motion sensor is connected to the central control chip. The motion sensor is used to sample the three-dimensional animation motion data in the TV game or computer game and transmit it to the central control chip. The three-dimensional animation motion data carries the three-dimensional acceleration data of the joystick in the TV game or computer game, the three-dimensional gyroscope data or the three-dimensional magnetometer data. The central control chip is used to process the collected three-dimensional animation motion data, that is, to finally obtain the movement distance data of the joystick in the TV game or computer game as the three-dimensional animation motion data resource, the angular displacement data of the joystick or the movement distance data of the joystick, thereby achieving the management and formation of the three-dimensional animation motion data resource.

[0004] On the other hand, the failure state of the joystick can reflect the failure state of the joystick according to the unreasonable state of the 3D animation motion data. There are many unexpected noise disturbances introduced by the environmental electric field in the environment where the joystick is located, which brings about noise values ​​​​mixed in the 3D animation motion data. Such noise values ​​​​often cause fluctuations in the 3D animation motion data, so that the 3D animation motion data contains noise values. In the face of the increasing demand for the usability and stability of joystick control in TV games or computer games, there is also a stronger demand for the usability of joystick control.

[0005] The current Z-score method is often used to obtain the Z score in 3D animation motion data, so as to obtain the unreasonable data in the 3D animation motion data. However, when the Z-score method is used to detect the 3D animation motion data, it will often be affected by noise values. The noise values ​​are often mistaken for data points of unreasonable data, or the data points of real unreasonable data are covered up, which is not conducive to the accuracy and usability of the Z-score method. Summary of the invention

[0006] In order to solve the defects in the prior art, the present invention proposes a 3D animation motion data resource management and generation system, which uses the Z-score method to perform unreasonable detection on the 3D animation motion data, obtains unreasonable data, and constructs an unreasonable data group; calculates the clutter value of the unreasonable data in the unreasonable data group and the clutter value of the unreasonable data in the near area to calibrate the Z score of the unreasonable data in the Z-score method, and obtains the characteristic Z score of each unreasonable data; if the characteristic Z score is higher than the score critical amount, it is confirmed that the target unreasonable data is correct unreasonable data, and if the characteristic Z score is higher than the score critical amount, it is confirmed that the target unreasonable data is correct unreasonable data. If the Z score is lower than the critical score, it is confirmed that the target unreasonable data is a noise value; after forming a correct data group based on the correct unreasonable data, it is sent to the pre-set restricted Boltzmann machine for advance estimation to learn, and obtain the advance estimation information of the three-dimensional animation motion data at the next point in time; if the advance estimation information of the three-dimensional animation motion data of the joystick at the next point in time is unreasonable, it is confirmed that the joystick is in a failure state; then the restricted Boltzmann machine for advance estimation is used to execute the advance estimation of the unreasonable data of the time point queue, and the queue information and unreasonable marking information are effectively used to accurately obtain the estimated unreasonable data in advance.

[0007] The present invention uses the following technical solutions.

[0008] A method for managing and generating three-dimensional animation motion data resources, comprising:

[0009] The motion sensor samples the 3D animation motion data in the TV game or computer game and transmits it to the central control chip. The central control chip performs pre-processing on the received 3D animation motion data and then performs processing on the pre-processed 3D animation motion data, that is, obtaining the 3D animation motion data resource in the TV game or computer game;

[0010] The method for the central control chip to pre-process the received three-dimensional animation motion data includes:

[0011] Step 1, using the Z-score method to perform unreasonable detection on the 3D animation motion data, obtain unreasonable data, and construct an unreasonable data group;

[0012] Step 2, calculate the clutter value of the unreasonable data in the unreasonable data group and the clutter value of the unreasonable data in the near area to calibrate the Z score of the unreasonable data in the Z-score method and obtain the characteristic Z score of each unreasonable data;

[0013] Step 3, if the characteristic Z score is higher than the score threshold, confirm that the target unreasonable data is correct unreasonable data; if the characteristic Z score is lower than the score threshold, confirm that the target unreasonable data is a noise value;

[0014] Step 4, after forming a correct data group based on the correct unreasonable data, send it to the pre-set restricted Boltzmann machine for advance estimation to learn, and obtain the advance estimation information of the three-dimensional animation motion data at the next point in time. If the advance estimation information of the three-dimensional animation motion data at the next point in time of the joystick is unreasonable, it is confirmed that the joystick is in a malfunction state.

[0015] Preferably, the motion sensor samples the 3D animation motion data of the joystick, and the sampling frequency of the 3D animation motion data is set to once per second.

[0016] Preferably, step 1 specifically includes: setting the number of three-dimensional animation motion data within the corresponding near zone L of each three-dimensional animation motion data to sixty, obtaining the Z score of each three-dimensional animation motion data, and registering the three-dimensional animation motion data with a Z score higher than one as unreasonable data, and the near zone L is the three-dimensional animation motion data with the highest Pearson coefficient of the corresponding three-dimensional animation motion data among the top sixty.

[0017] Preferably, step 2 specifically includes:

[0018] Step 2-1, based on each unreasonable data and the average of all 3D animation motion data in its corresponding near area, plus the difference between the unreasonable data and a pair of 3D animation motion data that are adjacent to each other at the sampling time point, obtain the partial variation of the unreasonable data, take the partial variation as the adjustment coefficient, and calculate the noise value of each unreasonable data;

[0019] Step 2-2, taking any unreasonable data in the unreasonable data group as the target unreasonable data, performing time point queue advance estimation on all 3D animation motion data in the near area of ​​the target unreasonable data, obtaining the advance estimated value of each 3D animation motion data in the near area, subtracting the advance estimated value of the 3D animation motion data from the true value to obtain the error value of the 3D animation motion data, calculating the error values ​​of all 3D animation motion data in the near area, performing cumulative summation, and obtaining the error representation of the near area of ​​the target unreasonable data;

[0020] Step 2-3, calculate the average of the KL divergence between the unreasonable target data and the corresponding three-dimensional animation motion data in the near zone, take the average as the density amplitude of the unreasonable target data, calculate the error manifestation of the near zone of the unreasonable target data multiplied by the density amplitude, and take the quantity as the clutter value manifestation of the near zone of the unreasonable target data;

[0021] Step 2-4, based on the clutter value of the unreasonable data and the clutter value of the corresponding near area, calibrate the Z score of the unreasonable data to obtain the characteristic Z score of each unreasonable data.

[0022] Preferably, in step 2-1, the operation equation of the partial variation is:

[0023] Δz j =|z j -z ju-1 |+|z j +z ju+1 |

[0024] The value of the overall animation motion data, z ju-1 represents the value of the 3D animation motion data sampled at the previous time point of the sampling time point of the jth unreasonable data, z ju+1 Represents the value of the 3D animation motion data sampled at a time point after the sampling time point of the j-th unreasonable data.

[0025] Preferably, in step 2-1, the computational equation for the clutter value of the unreasonable data is:

[0026]

[0027] In the equation, A j Represents the noise value of the jth unreasonable data, z j represents the value of the jth unreasonable 3D animation motion data, ν l represents the jth unreasonable data and the average of all 3D animation motion data in the corresponding near area, represents the variance of the jth unreasonable data and all 3D animation motion data in its corresponding nearby area, Δz j Represents the partial change of the j-th unreasonable data.

[0028] Preferably, in step 2-3, the computational equation for the clutter value in the near area of ​​the target unreasonable data is:

[0029]

[0030] In the equation, A jl The clutter value of the area near the jth unreasonable data is reflected, z jl,k represents the value of the kth 3D animation motion data in the lth near area of ​​the jth unreasonable data, z j represents the value of the jth unreasonable 3D animation motion data, u jl,k represents the sampling time point of the kth 3D animation motion in the l-near region of the jth unreasonable data, u j is the sampling time point of the jth unreasonable 3D animation motion data, represents the advance estimate of the kth 3D animation motion data in the lth near area of ​​the jth unreasonable data, NBQF jl Represents the root mean square error of the advance estimate of the 3D animation motion data in the near area of ​​the jth unreasonable data.

[0031] Preferably, in step 2-4, the calculation equation of the characteristic Z score is:

[0032]

[0033] In the equation, Represents the characteristic Z score of the jth unreasonable data, mpg j represents the starting Z score of the jth unreasonable data, A j Represents the noise value of the jth unreasonable data, A jl It represents the clutter value of the area near the j-th unreasonable data, zg(z) represents the highest amount of all unreasonable data in the unreasonable data group, and zd(z) represents the lowest amount of all unreasonable data in the unreasonable data group.

[0034] Preferably, in step 3, the fractional threshold is set to one.

[0035] Preferably, in step 4, the method of sending the correct data group into a pre-set restricted Boltzmann machine for advance estimation for learning includes: treating the unreasonable data group as a learning sample, sending it into the restricted Boltzmann machine for advance estimation, obtaining the output value, and using the KL divergence cost function to calculate the cost value of the restricted Boltzmann machine for advance estimation based on the output value and the true value of the three-dimensional animation motion data, repeatedly refreshing the parameters of the restricted Boltzmann machine for advance estimation until the cost value is lower than the set cost value, terminating the refresh, and obtaining the learned restricted Boltzmann machine for advance estimation.

[0036] Preferably, in step 4, the identified clutter values ​​are removed to obtain pre-processed 3D animation motion data, and the correct unreasonable data are formed into a correct data group;

[0037] If the advance estimation information of the 3D animation motion data at a later time point of the joystick is reasonable, it is confirmed that the joystick is in a reasonable state.

[0038] A 3D animation motion data resource management and generation system, comprising:

[0039] The motion sensor is connected to the central control chip. The motion sensor is used to sample the 3D animation motion data in the TV game or computer game and transmit it to the central control chip. The central control chip is used to perform pre-processing on the received 3D animation motion data and then perform processing on the pre-processed 3D animation motion data, that is, to obtain the 3D animation motion data resources in the TV game or computer game.

[0040] The modules running on the central control chip include:

[0041] A detection module, which is used to perform unreasonable detection on the three-dimensional animation motion data using a Z-score method, obtain unreasonable data, and construct an unreasonable data group;

[0042] A representation module is used to calculate the clutter value representation of unreasonable data in an unreasonable data group and the clutter value representation of unreasonable data in a near area, so as to calibrate the Z score of unreasonable data in a Z-score method and obtain the characteristic Z score of each unreasonable data;

[0043] A confirmation module, which is used to confirm that the target unreasonable data is correct unreasonable data if the characteristic Z score is higher than the score threshold, and to confirm that the target unreasonable data is a noise value if the characteristic Z score is lower than the score threshold;

[0044] The estimation module is used to form a correct data group based on the correct unreasonable data and then send it to a pre-set restricted Boltzmann machine for advance estimation to learn, so as to obtain the advance estimation information of the three-dimensional animation motion data at a later time point. If the advance estimation information of the three-dimensional animation motion data at a later time point of the joystick is unreasonable, it is confirmed that the joystick is in a malfunction state.

[0045] The beneficial effects of the present invention are that, compared with the prior art, the technical effects of the present invention include:

[0046] The Z-score method is used to perform unreasonable detection on the three-dimensional animation motion data, obtain unreasonable data, and construct an unreasonable data group; the clutter value manifestation of the unreasonable data in the unreasonable data group and the clutter value manifestation of the unreasonable data in the near area are calculated to calibrate the Z score of the unreasonable data in the Z-score method, and obtain the characteristic Z score of each unreasonable data; if the characteristic Z score is higher than the score critical value, it is confirmed that the target unreasonable data is the correct unreasonable data, and if the characteristic Z score is lower than the score critical value, it is confirmed that the target unreasonable data is a clutter value; after forming a correct data group based on the correct unreasonable data, it is sent to a pre-set restricted Boltzmann machine for advance estimation to learn, and the advance estimation information of the three-dimensional animation motion data at a later time point is obtained. If the advance estimation information of the three-dimensional animation motion data of the joystick at a later time point is unreasonable, it is confirmed that the joystick is in a malfunction state; then the restricted Boltzmann machine for advance estimation is used to perform the advance estimation of the unreasonable data of the time point queue, and the unreasonable data estimated in advance is accurately obtained through the queue information and the unreasonable marking information. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a flow chart of the method for managing and generating three-dimensional animation motion data resources in the present invention;

[0048] Figure 2 It is a partial structural diagram of the three-dimensional animation motion data resource management and generation system described in the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present invention clearer, the following will be combined with the drawings in the embodiments of the present invention to clearly and completely express the technical solution of the present invention. The embodiments expressed in the present invention are only some embodiments of the present invention, not all embodiments. According to the spirit of the present invention, other embodiments obtained by ordinary technicians in this field without creative work are all within the protection scope of the present invention.

[0050] like Figure 1 As shown, a 3D animation motion data resource management and generation method according to the present invention comprises:

[0051] The motion sensor samples the 3D animation motion data in the TV game or computer game and transmits it to the central control chip. The 3D animation motion data carries the 3D acceleration data, 3D gyroscope data or 3D magnetometer data of the joystick in the TV game or computer game. The 3D acceleration data, 3D gyroscope data or 3D magnetometer data of the joystick are the 3D motion data of the joystick. The central control chip performs preprocessing on the collected 3D animation motion data, and then performs processing on the preprocessed 3D animation motion data, that is, finally obtains the moving distance data, angular displacement data or moving distance data of the joystick in the TV game or computer game as the 3D animation motion data resource, thereby achieving the management and formation of the 3D animation motion data resource;

[0052] The method for the central control chip to pre-process the received three-dimensional animation motion data includes:

[0053] The present application reflects the failure state of the joystick based on unreasonable data in the three-dimensional animation motion data sampled from the joystick. The three-dimensional animation motion data can be any one of the three-dimensional acceleration data, three-dimensional gyroscope data or three-dimensional magnetometer data of the joystick in a TV game or a computer game.

[0054] Step 1, using the Z-score method to perform unreasonable detection on the 3D animation motion data, obtain unreasonable data, and construct an unreasonable data group;

[0055] In a preferred but non-limiting embodiment of the present invention, the motion sensor samples the three-dimensional animation motion data of the joystick, and the sampling frequency of the three-dimensional animation motion data is set to once per second.

[0056] According to the detection of the previous 3D animation motion data during the sampling period, because there are many unexpected noise disturbances introduced by the environmental electric field in the environment where the joystick is located, the sampled 3D animation motion data contains noise values. The noise values ​​are not conducive to the accuracy of the unreasonable data identified, which will also increase the misrecognition rate of the unreasonable data estimated in advance. Therefore, it is necessary to calibrate the Z score of the unreasonable data identified by the Z-score method to obtain the correct unreasonable data and reduce the misrecognition rate.

[0057] In a preferred but non-limiting implementation of the present invention, step 1 specifically includes: setting the number of three-dimensional animation motion data within the corresponding near zone L of each three-dimensional animation motion data to sixty, obtaining the Z score of each three-dimensional animation motion data, and registering the three-dimensional animation motion data with a Z score higher than one as unreasonable data. The near zone L is the three-dimensional animation motion data with the highest Pearson coefficient of the corresponding three-dimensional animation motion data for the top sixty.

[0058] Step 2, calculate the clutter value of the unreasonable data in the unreasonable data group and the clutter value of the unreasonable data in the near area to calibrate the Z score of the unreasonable data in the Z-score method and obtain the characteristic Z score of each unreasonable data;

[0059] In a preferred but non-limiting embodiment of the present invention, step 2 specifically comprises:

[0060] Step 2-1, based on each unreasonable data and the average of all 3D animation motion data in its corresponding near area, plus the difference between the unreasonable data and a pair of 3D animation motion data that are adjacent to each other at the sampling time point, obtain the partial variation of the unreasonable data, take the partial variation as the adjustment coefficient, and calculate the noise value of each unreasonable data;

[0061] In a preferred but non-limiting embodiment of the present invention, in step 2-1, the operation equation of the partial variation is:

[0062] Δz j =|z j -z ju-1 |+|z j +z ju+1 |

[0063] The value of the overall animation motion data, z ju-1 represents the value of the 3D animation motion data sampled at the previous time point of the sampling time point of the jth unreasonable data, z ju+1 Represents the value of the 3D animation motion data sampled at a time point after the sampling time point of the j-th unreasonable data.

[0064] The higher the difference between the target unreasonable data and the three-dimensional animation motion data sampled at a pair of consecutive time points adjacent to the sampling time point, the higher the partial variation of the target unreasonable data, and the more reliable the corresponding noise value performance.

[0065] In a preferred but non-limiting embodiment of the present invention, in step 2-1, the computational equation of the clutter value of the unreasonable data is:

[0066]

[0067] In the equation, A j Represents the noise value of the jth unreasonable data, z j represents the value of the jth unreasonable 3D animation motion data, ν l represents the jth unreasonable data and the average of all 3D animation motion data in the corresponding near area, represents the variance of the jth unreasonable data and all 3D animation motion data in its corresponding nearby area, Δz j Represents the partial change of the j-th unreasonable data.

[0068] and Represents the minimum limit of the clutter value. Represents the maximum limit of the clutter value. It represents the noise value when the unreasonable data exceeds the minimum limit of the noise value. Represents the clutter value when the unreasonable data exceeds the maximum clutter value limit, Δz j Used to reflect the confidence level of the clutter value.

[0069] In this application, It is used to confirm the dispersion amplitude and unreasonable state of 3D animation motion data, and reflect the clutter value of unreasonable data.

[0070] Add the unreasonable data of the target and the average of all 3D animation motion data in the near area As the maximum limit of the clutter value, the target unreasonable data and the average of all 3D animation motion data in the near area are subtracted Take it as the minimum limit of the clutter value; when the target unreasonable data exceeds the maximum limit of the clutter value, the higher the difference from the maximum limit of the clutter value, the higher the error between the target unreasonable data and all three-dimensional motion data in the near area, and the higher the clutter value performance; when the target unreasonable data exceeds the minimum limit of the clutter value, the higher the difference from the minimum limit of the clutter value, the higher the error between the target unreasonable data and all three-dimensional motion data in the near area, and the higher the clutter value performance.

[0071] Step 2-2, taking any unreasonable data in the unreasonable data group as the target unreasonable data, performing a time point queue (a time point queue is a queue formed by the three-dimensional animation motion data sampled at each sampling time point formed in the order of the sampling time points) on all three-dimensional animation motion data in the near area of ​​the target unreasonable data in advance, obtaining the advance estimated value of each three-dimensional animation motion data in the near area, subtracting the advance estimated value of the three-dimensional animation motion data from the true value to obtain the error value of the three-dimensional animation motion data, calculating the error values ​​of all three-dimensional animation motion data in the near area, performing cumulative summation, and obtaining the error manifestation of the near area of ​​the target unreasonable data;

[0072] In the present application, the time point queue advance estimation can be performed using the least squares method to obtain the advance estimation value of all three-dimensional animation motion data in the l near zone. The higher the impact of the noise value caused by the target unreasonable data, the higher the error amplitude of the error value, which is the difference between the true value and the advance estimation value. Therefore, the noise value of the near zone of the target unreasonable data can be reflected based on the cumulative sum of the error values ​​of other three-dimensional animation motion data in the l near zone of the target unreasonable data. The higher the density amplitude of the l near zone of the synchronized target unreasonable data, the higher the aggregation amplitude of the noise value, and the higher the impact on the Z score of the target unreasonable data.

[0073] Step 2-3, calculate the average of the KL divergence between the unreasonable target data and the corresponding three-dimensional animation motion data in the near zone, take the average as the density amplitude of the unreasonable target data, calculate the error manifestation of the near zone of the unreasonable target data multiplied by the density amplitude, and take the quantity as the clutter value manifestation of the near zone of the unreasonable target data;

[0074] In the present application, a pair of categories obtained according to the KL divergence between the three-dimensional animation motion data are the value of the three-dimensional animation motion data and the corresponding sampling time point. The higher the mean of the KL divergence is, the higher the density amplitude of the near zone is, and the higher the error manifestation of the near zone of the target unreasonable data is. The higher the density amplitude of the near zone is, the higher the contribution of the noise value caused by the Z score of the target unreasonable data in the near zone is.

[0075] In a preferred but non-limiting embodiment of the present invention, in step 2-3, the computational equation of the clutter value in the near area of ​​the target unreasonable data is:

[0076]

[0077] In the equation, A jl The clutter value of the area near the jth unreasonable data is reflected, z jl,k represents the value of the kth 3D animation motion data in the lth near area of ​​the jth unreasonable data, z jrepresents the value of the jth unreasonable 3D animation motion data, u jl,k represents the sampling time point of the kth 3D animation motion in the l-near region of the jth unreasonable data, u j is the sampling time point of the jth unreasonable 3D animation motion data, sixty represents the number of 3D animation motion data in the near area, represents the advance estimate of the kth 3D animation motion data in the lth near area of ​​the jth unreasonable data, NBQF jl Represents the root mean square error of the advance estimate of the 3D animation motion data in the near area of ​​the jth unreasonable data.

[0078] The dense amplitude in the near area representing unreasonable data, Error manifestation in the near area representing unreasonable data.

[0079] The higher the clutter value in the vicinity of the unreasonable data, or the higher the clutter value of the unreasonable data itself, the higher the contribution of the clutter value to the Z score of the unreasonable data. The Z score of the unreasonable data must be reduced according to the contribution of the clutter value to reduce the contribution of the clutter value to the Z score of the unreasonable data and obtain the Z score after calibration of the unreasonable data.

[0080] Step 2-4, based on the clutter value of the unreasonable data and the clutter value of the corresponding near area, calibrate the Z score of the unreasonable data to obtain the characteristic Z score of each unreasonable data.

[0081] In a preferred but non-limiting embodiment of the present invention, in step 2-4, the calculation equation of the characteristic Z score is:

[0082]

[0083] In the equation, Represents the characteristic Z score of the jth unreasonable data, mpg j represents the starting Z score of the jth unreasonable data, A j Represents the noise value of the jth unreasonable data, A jl It represents the clutter value of the area near the j-th unreasonable data, zg(z) represents the highest amount of all unreasonable data in the unreasonable data group, and zd(z) represents the lowest amount of all unreasonable data in the unreasonable data group.

[0084] And A j +A jl Represents the contribution of the noise value caused by the Z score of the current unreasonable 3D animation motion data, and zg(z)-zd(z) is used to perform normalization on the contribution of the noise value.

[0085] Step 3, if the characteristic Z score is higher than the score threshold, confirm that the target unreasonable data is correct unreasonable data; if the characteristic Z score is lower than the score threshold, confirm that the target unreasonable data is a noise value;

[0086] In a preferred but non-limiting embodiment of the present invention, in step 3, the fractional threshold is set to one.

[0087] Step 4, after forming a correct data group based on the correct unreasonable data, send it to the pre-set restricted Boltzmann machine for advance estimation to learn, and obtain the advance estimation information of the three-dimensional animation motion data at the next point in time. If the advance estimation information of the three-dimensional animation motion data at the next point in time of the joystick is unreasonable, it is confirmed that the joystick is in a malfunction state.

[0088] In a preferred but non-restrictive embodiment of the present invention, in step 4, the method of sending the correct data group into a pre-set restricted Boltzmann machine for advance estimation for learning includes: treating the unreasonable data group as a learning sample, sending it into the restricted Boltzmann machine for advance estimation, obtaining the output value, and using the KL divergence cost function to calculate the cost value of the restricted Boltzmann machine for advance estimation based on the output value and the true value of the three-dimensional animation motion data, repeatedly refreshing the parameters of the restricted Boltzmann machine for advance estimation until the cost value is lower than the set cost value, terminating the refresh, and obtaining the learned restricted Boltzmann machine for advance estimation.

[0089] In a preferred but non-limiting embodiment of the present invention, in step 4, the correct data group is the unreasonable data obtained by adjusting the characteristic Z score and the preset score threshold, and the adjusted characteristic Z score removes the identified clutter value to obtain the pre-processed three-dimensional animation motion data, and the correct unreasonable data forms the correct data group, which reduces the misrecognition state caused by the clutter value and improves the accuracy of the three-dimensional animation data of the joystick;

[0090] The present application reflects the malfunction state of the joystick based on the unreasonable state of the 3D animation motion data. If the advance estimation information of the 3D animation motion data at a later time point of the joystick is reasonable, it is confirmed that the joystick is in a reasonable state.

[0091] like Figure 2 As shown, a 3D animation motion data resource management and generation system of the present invention comprises:

[0092] The motion sensor is connected to the central control chip. The motion sensor is used to sample the 3D animation motion data in the TV game or computer game and transmit it to the central control chip. The 3D animation motion data carries the 3D acceleration data, 3D gyroscope data or 3D magnetometer data of the joystick in the TV game or computer game. The 3D acceleration data, 3D gyroscope data or 3D magnetometer data of the joystick are the 3D motion data of the joystick. The central control chip is used to perform preprocessing on the collected 3D animation motion data, and then perform processing on the preprocessed 3D animation motion data, that is, finally to obtain the moving distance data, angular displacement data or moving distance data of the joystick in the TV game or computer game as the 3D animation motion data resource, thereby achieving the management and formation of the 3D animation motion data resource;

[0093] The modules running on the central control chip include:

[0094] A detection module, which is used to perform unreasonable detection on the three-dimensional animation motion data using a Z-score method, obtain unreasonable data, and construct an unreasonable data group;

[0095] A representation module is used to calculate the clutter value representation of unreasonable data in an unreasonable data group and the clutter value representation of unreasonable data in a near area, so as to calibrate the Z score of unreasonable data in a Z-score method and obtain the characteristic Z score of each unreasonable data;

[0096] A confirmation module, which is used to confirm that the target unreasonable data is correct unreasonable data if the characteristic Z score is higher than the score threshold, and to confirm that the target unreasonable data is a noise value if the characteristic Z score is lower than the score threshold;

[0097] The estimation module is used to form a correct data group based on the correct unreasonable data and then send it to a pre-set restricted Boltzmann machine for advance estimation to learn, so as to obtain the advance estimation information of the three-dimensional animation motion data at a later time point. If the advance estimation information of the three-dimensional animation motion data at a later time point of the joystick is unreasonable, it is confirmed that the joystick is in a malfunction state.

[0098] The beneficial effects of the present invention are that, compared with the prior art, the technical effects of the present invention include:

[0099] The Z-score method is used to perform unreasonable detection on the three-dimensional animation motion data, obtain unreasonable data, and construct an unreasonable data group; the clutter value manifestation of the unreasonable data in the unreasonable data group and the clutter value manifestation of the unreasonable data in the near area are calculated to calibrate the Z score of the unreasonable data in the Z-score method, and obtain the characteristic Z score of each unreasonable data; if the characteristic Z score is higher than the score critical value, it is confirmed that the target unreasonable data is the correct unreasonable data, and if the characteristic Z score is lower than the score critical value, it is confirmed that the target unreasonable data is a clutter value; after forming a correct data group based on the correct unreasonable data, it is sent to a pre-set restricted Boltzmann machine for advance estimation to learn, and the advance estimation information of the three-dimensional animation motion data at a later time point is obtained. If the advance estimation information of the three-dimensional animation motion data of the joystick at a later time point is unreasonable, it is confirmed that the joystick is in a malfunction state; then the restricted Boltzmann machine for advance estimation is used to perform the advance estimation of the unreasonable data of the time point queue, and the unreasonable data estimated in advance is accurately obtained through the queue information and the unreasonable marking information.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not deviate from the spirit and scope of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A method for managing and generating 3D animation motion data resources, characterized in that: include: The motion sensor samples the 3D animation motion data in the TV game or computer game and transmits it to the central control chip. The central control chip performs pre-processing on the received 3D animation motion data and then performs processing on the pre-processed 3D animation motion data, that is, obtaining the 3D animation motion data resource in the TV game or computer game; The method for the central control chip to pre-process the received three-dimensional animation motion data includes: Step 1, using the Z-score method to perform unreasonable detection on the 3D animation motion data, obtain unreasonable data, and construct an unreasonable data group; Step 2, calculate the clutter value of the unreasonable data in the unreasonable data group and the clutter value of the unreasonable data in the near area to calibrate the Z score of the unreasonable data in the Z-score method and obtain the characteristic Z score of each unreasonable data; Step 3, if the characteristic Z score is higher than the score threshold, confirm that the target unreasonable data is correct unreasonable data; if the characteristic Z score is lower than the score threshold, confirm that the target unreasonable data is a noise value; Step 4, after forming a correct data group based on the correct unreasonable data, send it to the pre-set restricted Boltzmann machine for advance estimation to learn, and obtain the advance estimation information of the three-dimensional animation motion data at the next point in time. If the advance estimation information of the three-dimensional animation motion data at the next point in time of the joystick is unreasonable, it is confirmed that the joystick is in a malfunction state.

2. The method for managing and generating 3D animation motion data resources according to claim 1, characterized in that: The motion sensor samples the 3D animation motion data of the joystick, and the sampling frequency of the 3D animation motion data is set to once per second.

3. The method for managing and generating 3D animation motion data resources according to claim 2, characterized in that: Step 1 specifically includes: setting the number of 3D animation motion data within the corresponding near zone L of each 3D animation motion data to sixty, obtaining the Z score of each 3D animation motion data, and registering the 3D animation motion data with a Z score higher than one as unreasonable data. The near zone L is the 3D animation motion data with the highest Pearson coefficient of the corresponding 3D animation motion data among the top sixty.

4. The method for managing and generating 3D animation motion data resources according to claim 3, characterized in that: Step 2 specifically includes: Step 2-1, based on each unreasonable data and the average of all 3D animation motion data in its corresponding near area, plus the difference between the unreasonable data and a pair of 3D animation motion data that are adjacent to each other at the sampling time point, obtain the partial variation of the unreasonable data, take the partial variation as the adjustment coefficient, and calculate the noise value of each unreasonable data; Step 2-2, taking any unreasonable data in the unreasonable data group as the target unreasonable data, performing time point queue advance estimation on all 3D animation motion data in the near area of ​​the target unreasonable data, obtaining the advance estimated value of each 3D animation motion data in the near area, subtracting the advance estimated value of the 3D animation motion data from the true value to obtain the error value of the 3D animation motion data, calculating the error values ​​of all 3D animation motion data in the near area, performing cumulative summation, and obtaining the error representation of the near area of ​​the target unreasonable data; Step 2-3, calculate the average of the KL divergence between the unreasonable target data and the corresponding three-dimensional animation motion data in the near zone, take the average as the density amplitude of the unreasonable target data, calculate the error manifestation of the near zone of the unreasonable target data multiplied by the density amplitude, and take the quantity as the clutter value manifestation of the near zone of the unreasonable target data; Step 2-4, based on the clutter value of the unreasonable data and the clutter value of the corresponding near area, calibrate the Z score of the unreasonable data to obtain the characteristic Z score of each unreasonable data.

5. The method for managing and generating 3D animation motion data resources according to claim 4, characterized in that: In step 2-1, the calculation equation of the partial variation is: Δz j =|with j -with ju-1 |+|with j +with ju+1 | The value of the overall animation motion data, z ju-1 represents the value of the 3D animation motion data sampled at the previous time point of the sampling time point of the jth unreasonable data, z ju+1 represents the value of the 3D animation motion data sampled at a time point after the sampling time point of the j-th unreasonable data; In step 2-1, the calculation equation of the clutter value of unreasonable data is: In the equation, A j Represents the noise value of the jth unreasonable data, z j represents the value of the jth unreasonable 3D animation motion data, ν l represents the jth unreasonable data and the average of all 3D animation motion data in the corresponding near area, represents the variance of the jth unreasonable data and all 3D animation motion data in its corresponding nearby area, Δz j Represents the partial change of the j-th unreasonable data.

6. The method for managing and generating 3D animation motion data resources according to claim 5, characterized in that: In step 2-3, the calculation equation of the clutter value in the near area of ​​the target unreasonable data is: In the equation, A jl The clutter value of the area near the jth unreasonable data is reflected, z jl,k represents the value of the kth 3D animation motion data in the lth near area of ​​the jth unreasonable data, z j represents the value of the jth unreasonable 3D animation motion data, u jl,k represents the sampling time point of the kth 3D animation motion in the l-near region of the jth unreasonable data, u j is the sampling time point of the jth unreasonable 3D animation motion data, represents the advance estimate of the kth 3D animation motion data in the lth near area of ​​the jth unreasonable data, NBQF jl Represents the root mean square error of the advance estimate of the 3D animation motion data in the near area of ​​the jth unreasonable data.

7. The method for managing and generating 3D animation motion data resources according to claim 6, characterized in that: In steps 2-4, the characteristic Z score is calculated as: In the equation, Represents the characteristic Z score of the jth unreasonable data, mpg j represents the starting Z score of the jth unreasonable data, A j Represents the noise value of the jth unreasonable data, A jl It represents the clutter value of the area near the j-th unreasonable data, zg(z) represents the highest amount of all unreasonable data in the unreasonable data group, and zd(z) represents the lowest amount of all unreasonable data in the unreasonable data group.

8. The method for managing and generating 3D animation motion data resources according to claim 7, characterized in that: In step 3, the fractional threshold is set to one.

9. The method for managing and generating 3D animation motion data resources according to claim 8, characterized in that: In step 4, the method of sending the correct data set into a pre-set restricted Boltzmann machine for advance estimation for learning includes: taking the unreasonable data set as a learning sample, sending it into the restricted Boltzmann machine for advance estimation, obtaining the output value, using the KL divergence cost function to calculate the cost value of the restricted Boltzmann machine for advance estimation based on the output value and the true value of the three-dimensional animation motion data, repeatedly refreshing the parameters of the restricted Boltzmann machine for advance estimation until the cost value is lower than the set cost value, terminating the refreshing, and obtaining the learned restricted Boltzmann machine for advance estimation; In step 4, the identified clutter values ​​are removed to obtain pre-processed 3D animation motion data, and the correct unreasonable data are formed into a correct data group; If the advance estimation information of the 3D animation motion data at a later time point of the joystick is reasonable, it is confirmed that the joystick is in a reasonable state.

10. A 3D animation motion data resource management and generation system, characterized in that: include: The motion sensor is connected to the central control chip. The motion sensor is used to sample the 3D animation motion data in the TV game or computer game and transmit it to the central control chip. The central control chip is used to perform pre-processing on the received 3D animation motion data and then perform processing on the pre-processed 3D animation motion data, that is, to obtain the 3D animation motion data resources in the TV game or computer game. The modules running on the central control chip include: A detection module, which is used to perform unreasonable detection on the three-dimensional animation motion data using a Z-score method, obtain unreasonable data, and construct an unreasonable data group; A representation module is used to calculate the clutter value representation of unreasonable data in an unreasonable data group and the clutter value representation of unreasonable data in a near area, so as to calibrate the Z score of unreasonable data in a Z-score method and obtain the characteristic Z score of each unreasonable data; A confirmation module, which is used to confirm that the target unreasonable data is correct unreasonable data if the characteristic Z score is higher than the score threshold, and to confirm that the target unreasonable data is a noise value if the characteristic Z score is lower than the score threshold; The estimation module is used to form a correct data group based on the correct unreasonable data and then send it to a pre-set restricted Boltzmann machine for advance estimation to learn, so as to obtain the advance estimation information of the three-dimensional animation motion data at a later time point. If the advance estimation information of the three-dimensional animation motion data at a later time point of the joystick is unreasonable, it is confirmed that the joystick is in a malfunction state.

Images