JP2014115943A - Animation content generation device, animation content generation method, and computer program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To generate animation content with a specific expression reflected, and further with improved robustness and adaptation to a noncyclical motion.SOLUTION: An animation content generation device includes: a leaning part 3 for using skeleton motion data of a learning-target specific expression and neutral skeleton motion data to calculate a parameter of a conversion function for converting the neutral skeleton motion data into the skeleton motion data of the specific expression in a primary component space of the skeleton motion data for each category of motion; a storage part 5 for storing the calculated parameter for each category of motion; an input part 2 for inputting neutral skeleton motion data of a conversion-target motion category; and a conversion part 4 for using the calculated parameter corresponding to the conversion-target motion category to convert the input neutral skeleton motion data into skeleton motion data of a specified specific expression in the primary component space by using the conversion function.

Description

  The present invention relates to a moving image content generation device, a moving image content generation method, and a computer program.

  In recent years, voice interactive user interfaces that run on mobile terminals have been commercialized, but in the future, not only conversation content but also video content that can be expressed more richly, such as displaying characters called "embodied conversational agent" It is thought that the need for will increase. Also, in the dialogue system, the visual that reflects the emotional expression that fosters the cuteness and familiarity of the character is an important element for improving the user's satisfaction, and the technology to generate video content that reflects the emotional expression is It becomes important.

  As content data used for displaying moving image content by computer graphics, for example, skeleton type motion data (hereinafter referred to as skeleton motion data) is known. Here, the skeleton motion data of the human body exemplified in FIG. 5 will be described as data representing the motion of the posture (pose) of the human body. FIG. 5 is a schematic diagram of a definition example of human skeleton motion data. Skeleton motion data is based on the human skeleton, using bone and bone connection points (joints), with one joint as the root (root), and the structure of bones that are sequentially connected from the root via the joint (tree) Defined as a structure. FIG. 5 shows only a part of the definition of the skeleton motion data. In FIG. 5, the joint 100 is a waist part and is defined as a root. Joint 101 is the elbow portion of the left arm, Joint 102 is the wrist portion of the left arm, Joint 103 is the elbow portion of the right arm, Joint 104 is the wrist portion of the right arm, Joint 105 is the knee portion of the left foot, and Joint 106 is the left foot portion. Ankle part, joint 107 is right leg knee part, joint 108 is right leg ankle part, joint 109 is clavicle part, joints 110 and 111 are shoulder parts, joint 112 is neck part, joints 113 and 114 are The hip joint part.

  The skeleton movement data is data that records the movement of each joint of the skeleton type object, and a human body, an animal, a robot, or the like is applicable as the skeleton type object. As the skeleton motion data, position information, angle information, speed information, acceleration information, and the like of each joint can be used. Here, the human skeleton angle information will be described as an example of the skeleton motion data.

  Human body skeleton-type angle information data represents a series of movements of a person by a series of multiple poses. Basic pose data that represents a person's basic pose and each pose in the actual movement of the person. Frame data for each pose. The basic pose data includes information such as the position of the root and the position of each joint in the basic pose, and the length of each bone. The basic pose is specified by the basic pose data. The frame data represents the amount of movement from the basic pose for each joint. Here, angle information is used as the movement amount. Each frame data identifies each pose in which each movement amount is added to the basic pose. Thereby, a series of movements of a person is specified by the continuation of each pose specified by each frame data. The human skeleton-type angle information data can be created by a motion capture process from an image obtained by photographing a person's movement with a camera, or can be created manually by key frame animation.

  Conventionally, a model has been created that learns the difference between a neutral movement that does not reflect emotional expression and a movement that reflects a specific emotional expression, and this model is used to change the input neutral movement. Techniques for converting are known. For example, in the prior art of Non-Patent Document 1, the speed and space width are learned for an “End Effector joint” such as a wrist. In the prior art of Non-Patent Document 2, a model approximated by a sine function is learned in the principal component space.

AMAYA, K., BRUDERLIN, A., AND CALVERT, T. 1996. Emotion from Motion. In Graphics Interface, 222-229. TROJE, N. 2002. Decomposing biological motion: A framework for analysis and synthesis of human gait patterns. In Journal of Vision. 2 (5): 371-387.

  However, since the above-described conventional technique of Non-Patent Document 1 does not consider the correlation between the joints, the correspondence between the neutral movement and the movement reflecting the specific emotion expression seems to have robustness in the joint space. It is difficult to ask for. Further, the conventional technique of Non-Patent Document 2 can cope with periodic movements such as walking, but is not suitable for non-periodic movements such as bowing. In general, the movement of the user in the dialogue system is often aperiodic, and the conventional technology of Non-Patent Document 2 is insufficient.

  The present invention has been made in consideration of such circumstances, and when generating moving image content reflecting a specific expression, the moving image content generating device can be adapted to improved robustness and aperiodic movement. It is an object to provide a moving image content generation method and a computer program.

  In order to solve the above-described problem, the video content generation device according to the present invention uses, for each type of motion, the skeleton motion data of the skeleton motion data using the skeleton motion data of the specific expression to be learned and the neutral skeleton motion data. A learning unit that calculates parameters of a conversion function for converting the neutral skeleton motion data into the skeleton motion data of the specific expression in a component space, and a storage unit that stores the calculated parameters for each type of motion An input unit that inputs neutral skeleton motion data of the type of motion to be converted, and the input neutral by the conversion function using the calculated parameter corresponding to the type of motion to be converted Skeleton motion data of the specified representation in the principal component space Characterized by comprising a conversion unit for converting.

  In the moving image content generation device according to the present invention, the learning unit obtains a spatial parameter and a time parameter by a linear regression method for each segment in the principal component space of the skeleton motion data.

  In the moving image content generation device according to the present invention, the learning unit uses a differential value of principal component coordinates when calculating the spatial parameter by a linear regression method, and the conversion unit uses the spatial parameter to perform the conversion. A function is used to convert the differential value of the principal component coordinates of the input neutral skeleton motion data, and integrate the converted differential value.

  The moving image content generation method according to the present invention uses the skeleton motion data of a specific expression to be learned and the neutral skeleton motion data for each type of motion, and the neutral skeleton motion in the principal component space of the skeleton motion data. A learning step for calculating a parameter of a conversion function for converting data into skeleton motion data of the specific expression, a storage step for storing the calculated parameter for each type of motion, and a type of motion to be converted Using the input step of inputting neutral skeleton motion data and the calculated parameter corresponding to the type of motion to be converted, the input neutral skeleton motion data is converted into a principal component space by the conversion function. , Convert to skeleton motion data of specified specific representation Characterized by comprising a conversion step, the.

  The computer program according to the present invention uses, for each type of motion, the skeleton motion data of a specific expression to be learned and the neutral skeleton motion data to generate the neutral skeleton motion data in the principal component space of the skeleton motion data. A learning step for calculating a parameter of a conversion function for converting into the skeleton motion data of the specific expression, a storage step for storing the calculated parameter for each type of motion, and a neutral type of the type of motion to be converted Specifying the input neutral skeleton motion data in the principal component space by the conversion function using the input step of inputting skeleton motion data and the calculated parameter corresponding to the type of motion to be converted The skeleton motion data of the specified specific representation Characterized in that it is a computer program for executing a conversion step, to the computer.

  According to the present invention, it is possible to improve the robustness and adapt to non-periodic movement when generating moving image content reflecting a specific expression.

It is explanatory drawing for demonstrating the concept of this invention. It is a block diagram which shows the structure of the moving image content production | generation apparatus 1 which concerns on one Embodiment of this invention. It is a graph which shows the example of a production | generation of a segment. It is a graph which shows the example of normalization of a time axis. It is the schematic of the example of a definition of the skeleton motion data of a human body.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  First, the concept of the present invention will be described with reference to FIG. The present invention has a learning step S1 for learning a conversion function using learning data, and a conversion step S2 for converting a motion expression using the conversion function. In the learning stage S1, the conversion function is learned using neutral motion data that does not particularly reflect emotional expressions and motion data that reflects specific emotional expressions. In the example of FIG. 1, learning of the conversion function f1 is performed using the neutral motion data and the motion data reflecting the emotion expression 1, and the motion data reflecting the neutral motion data and the emotion expression 2 Is used to learn the conversion function f2. The conversion function f1 is a conversion function for reflecting the emotional expression 1 on the neutral motion data. The conversion function f2 is a conversion function for reflecting the emotion expression 2 on the neutral motion data.

  Next, in the conversion step S2, the motion expression is converted using a conversion function for reflecting the specified expression with respect to the neutral motion data. In the example of FIG. 1, the motion expression is converted to the neutral motion data by using the conversion function f1 for reflecting the designated emotion expression 1, and motion data reflecting the emotion expression 1 is generated. Has been. Also, the motion expression is converted to the neutral motion data using the conversion function f2 for reflecting the designated emotion expression 2, and the motion data reflecting the emotion expression 2 is generated.

  FIG. 2 is a block diagram showing a configuration of the moving image content generation device 1 according to the embodiment of the present invention. In FIG. 2, the moving image content generation device 1 includes an input unit 2, a learning unit 3, a conversion unit 4, a storage unit 5, and an output unit 6.

[Input section]
The input unit 2 inputs skeleton motion data for learning and skeleton motion data for conversion. The input unit 2 outputs the skeleton motion data for learning to the learning unit 3 and outputs the skeleton motion data for conversion to the conversion unit 4. The skeleton motion data for learning and the skeleton motion data for conversion are distinguished by the type of motion. There are various types of movements, such as walking movements, running movements, bouncing movements, crawl movements, swinging movements, and knocking movements.

  As the skeleton motion data for learning, skeleton motion data of a specific expression and neutral skeleton motion data are input for each type of motion. There are various types of expression such as emotion expression, gender expression, age expression, and local expression. There are various expressions of emotions such as laughing expressions, angry expressions, lonely expressions, depressed expressions, and happy expressions. Neutral skeleton motion data is skeleton motion data that does not particularly reflect the expression.

  As the skeleton motion data for conversion, neutral skeleton motion data is input for each type of motion.

  The input unit 2 also inputs expression designation data that designates what kind of expression is to be reflected in the skeleton motion data for conversion. This expression designation data is output to the conversion unit 4.

[Learning Department]
As shown in FIG. 2, the learning unit 3 includes a physical quantity conversion unit 11, a principal component analysis unit 12, a segment generation unit 13, a time axis normalization unit 14, a space axis linear regression unit 15, and a time axis linear regression unit 16. Have. For each type of motion, the learning unit 3 uses the skeleton motion data of a specific expression and the neutral skeleton motion data, and performs the linear skeleton motion data for each segment in the principal component space of the skeleton motion data by a linear regression method. Is converted into a skeleton motion data of the specific expression. Hereinafter, each part 11-16 is demonstrated in detail. The learning unit 3 receives skeleton motion data of a specific expression and neutral skeleton motion data for each type of motion from the input unit 2, but is simply referred to as skeleton motion data unless otherwise distinguished. .

[Physical quantity converter]
The physical quantity converter 11 generates data to be input to the principal component analyzer 12 from the input skeleton motion data. The physical quantity conversion unit 11 calculates at which position each joint is moving with respect to the root during the period of motion represented by the skeleton motion data. Here, the joint relative position at time t is calculated using human body skeleton-type angle information data which is a kind of skeleton motion data. The joint relative position is a relative position of the joint with respect to the root.

First, the physical quantity converter 11 calculates the joint position using the basic pose data and the frame data in the human body skeleton type angle information data. The basic pose data includes information for specifying the basic pose, such as the position of the root and the position of each joint in the basic pose, and the length of each bone. The frame data has information on the amount of movement from the basic pose for each joint. Here, angle information is used as the movement amount. In this case, the position p ^k (t) of the k-th joint at time t is calculated by the equations (1) and (2). p ^k (t) is represented by three-dimensional coordinates. Note that time t is the time of the frame data. In the present embodiment, time t is simply handled as a “frame index”. Thereby, the time t takes values of 0, 1, 2,..., T−1. T is the number of frames included in the human skeleton-type angle information data.

However, the 0th (i = 0) joint is the root. R _axis ^{i-1, i} (t) is a coordinate rotation matrix between the i-th joint and its parent joint ("i-1" -th joint), and is included in the basic pose data. A local coordinate system is defined for each joint, and the coordinate rotation matrix represents the correspondence of the local coordinate system between joints in a parent-child relationship. R ⁱ (t) is a rotation matrix of the i-th joint in the local coordinate system of the i-th joint, and is angle information included in the frame data. T ⁱ (t) is a transition matrix between the i-th joint and its parent joint, and is included in the basic pose data. The transition matrix represents the bone length between the i-th joint and its parent joint.

Next, the physical quantity conversion unit 11 calculates the relative position (joint relative position) p ′ ^k (t) of the k-th joint with respect to the root at time t using Equation (3).

Here, p ^root (t) is the position (p ⁰ (t)) of the route (0th joint) at time t.

Thus, the frame “x (t)” at time t is expressed as “x (t) = {p ′ ¹ (t), p ′ ² (t),..., P ′ ^K (t)} ^T ”. expressed. K is the number of joints excluding the root. ^T represents a transposed matrix.

[Principal component analysis section]
The principal component analysis unit 12 performs principal component analysis processing on the joint relative position data generated by the physical quantity conversion unit 11. Here, for a certain motion type S (for example, knocking motion), the joint relative position data X ⁰ _S regarding the neutral skeleton motion data and the joint regarding the skeleton motion data of a certain expression (for example, angry expression). set composed of the relative position data ^X _{1 S ^{'X = {X 0 S1, X}} 0 S2, ···, X 0 Ss, X 1 S1, X 1 S2, ···, X 1 Ss} a, S is an index of skeleton motion data, and one joint relative position data “X ^e _S ” is ^converted to “X ^e by using the frame“ x (t) ”at time t. _S = {x (t1), x (t2),..., X (tL)} where e is 0 or 1. L is a frame included in one joint relative position data. The total number of frames of all joint relative position data included in the set X is N. Accordingly, X is represented by a matrix of M rows and N columns (where M = 3 × K).

  In the principal component analysis process, the principal component analysis process is performed on the matrix X of M rows and N columns, and X is converted into the principal component space. For the principal component analysis processing, for example, “open source” is published on the Internet <URL: http://en.wikipedia.org/wiki/Principal_components_analysis#Software.2Fsource_code> and can be used.

Here, the principal component analysis method will be described.
First, the matrix D of N rows and M columns obtained by subtracting the average value from X is calculated by the equation (4).

  Next, singular value decomposition (Singular Value Decomposition) is performed on the matrix D of N rows and M columns according to the equation (5). As for the singular value decomposition process, for example, “open source” is published on the Internet <URL: http://www.gnu.org/software/gsl/> and can be used.

  However, U is a unitary matrix of N rows and N columns. Σ is a diagonal matrix having non-negative diagonal elements of N rows and M columns in descending order, and represents the variance of the coordinates of the principal component space. V is a unitary matrix of M rows and M columns, and is a coefficient (principal component) for the principal component.

  Next, the matrix D of N rows and M columns is converted into the principal component space by the equation (6). The matrix Y with M rows and N columns represents the coordinates of the principal component space.

  In the principal component analysis processing, a matrix (principal component coordinate matrix) Y representing coordinates of the principal component space and a coefficient matrix (principal component coefficient matrix) V for the principal components are stored in the storage unit 5.

  Note that the matrix X representing the coordinates of the original space and the principal component coordinate matrix Y can be converted into each other by the equations (6) and (7).

  Moreover, it can convert by (8) Formula by the upper r main components.

Here, V ^r is a matrix of M rows and r columns composed of upper r rows in the principal component coefficient matrix V. Y ^r is an r-row N-column matrix composed of the upper r columns in the principal component coordinate matrix Y. X ^~ is a matrix of reconstructed M rows and N columns.

The coordinate Y ^{r of the} principal component space corresponds to each skeleton motion data and is expressed as “Y ^r = {Y ^re _S1 , Y ^re _S2 ,..., Y ^re _Ss }. restored ^{X ~} correspond to the skeleton motion data after restoration, _{^{"X ~ = {X ~e S1,}} X ~e S2, ···, X ~e Ss} denoted.

[Segment generator]
The segment generation unit 13 generates a segment using one principal component coordinate ^Yr for each skeleton motion data in the principal component coordinate matrix Y calculated by the principal component analysis unit 12. For example, using the second principal component coordinate ^{Y 2.}

As a method of generating a segment for one skeleton motion data obtains an extreme value of the principal component coordinate Y ^r about the skeleton motion data, the boundary of the segment a frame of polar values. The first frame and the last frame related to the skeleton motion data are also segment boundaries. However, when the length of one segment is too short (when the length of one segment is less than the prescribed length), the rear boundary is deleted and the segment is lengthened. FIG. 3 is a graph showing an example of segment generation. Example of FIG. 3 relates skeleton motion data of the motion to knock, the second principal component coordinate Y ² is graphed about five skeleton motion data. In each graph in FIG. 3, a circle is a segment boundary. The segment generated by the segment generation unit 13 for the skeleton motion data is common to all principal component coordinates in the skeleton motion data.

[Time axis normalization section]
The time axis normalization unit 14 normalizes the time axis for the segment generated by the segment generation unit 13 for each skeleton motion data. The normalization of the time axis is performed for each skeleton motion data from the first principal component coordinate to the r-th principal component coordinate. As a normalization method of the time axis with respect to one principal component coordinate, for each principal component coordinate, the value of the principal component coordinate is set so that the number of frames in the segment becomes a prescribed frame number (for example, 100 frames). Interpolate. For this interpolation processing, for example, a technique called “cubic spline interpolation” is known, and “open source” is published on the Internet <URL: http://en.wikipedia.org/wiki/Spline_interpolation>. It is available.

For each segment, the ratio (interpolation ratio) between the number of frames before interpolation and the prescribed number of frames is stored in the storage unit 5. FIG. 4 is a graph showing an example of time axis normalization. The example of FIG. 4 relates to the skeleton motion data of the knocking motion, and the second principal component coordinates Y ² after normalization of the time axis for each of the skeleton motion data of the neutral and the skeleton motion data of the angry expression. Is graphed. By normalizing the time axis, the number of frames in each segment from the first principal component coordinate to the r-th principal component coordinate becomes the same for all the skeleton motion data.

[Spatial axis linear regression section]
The spatial axis linear regression unit 15 calculates a spatial parameter of a conversion function for converting neutral skeleton motion data into skeleton motion data of a specific expression by a linear regression method for each segment. At this time, the space axis linear regression unit 15 performs linear regression calculation according to equation (9) using the differential value of the principal component coordinates. This corresponds to the process of performing linear regression operation using the differential value of the principal component coordinates and then integrating in order to smooth the connection between segments in the spatial axis conversion in the conversion unit 4 described later. It is.

Here, q ⁱ (t) is a differential value of the j-th principal component coordinate in the t-th frame of the i-th segment with respect to the skeleton motion data of a specific expression. k ⁱ (t) is a differential value of the j-th principal component coordinate in the t-th frame of the i-th segment with respect to the neutral skeleton motion data.

The spatial axis linear regression unit 15 calculates the spatial parameters a ⁱ and b ⁱ related to the i-th segment by linear regression calculation using the above equation (9). Spatial parameters a ⁱ and b ⁱ are stored in the storage unit 5.

[Time-axis linear regression section]
The time axis linear regression unit 16 calculates a time parameter of a conversion function for converting neutral skeleton motion data into skeleton motion data of a specific expression by a linear regression method for each segment. At this time, the time-axis linear regression unit 16 performs linear regression calculation according to equation (10) using the interpolation ratio of each segment calculated by the time-axis normalization unit 14 and stored in the storage unit 5.

Here, t ¹ _i is the interpolation ratio of the i-th segment regarding the skeleton motion data of a specific expression. t ⁰ _i is the interpolation ratio of the i-th segment with respect to the neutral skeleton motion data.

The time axis linear regression unit 16 calculates a time parameter m ^ _i related to the i-th segment by linear regression calculation using the above equation (10). The time parameter m ^ _i is stored in the storage unit 5.

The above is the description of the learning unit 3 according to the present embodiment. The learning unit 3, the storage unit 5, each type of movement, and, for each specific expression, spatial parameters a ⁱ and b ⁱ and time parameters m ^ _i for each segment i is stored.

[Conversion section]
As shown in FIG. 2, the conversion unit 4 includes a physical quantity conversion unit 11, a principal component analysis unit 12, a segment generation unit 13, a time axis normalization unit 14, a space axis conversion unit 21, a time axis conversion unit 22, and a position space. A restoration unit 23 is included. The conversion unit 4 receives skeleton motion data and expression designation data for conversion from the input unit 2. As the skeleton motion data for conversion, neutral skeleton motion data is input for each type of motion. The expression designation data indicates an expression to be reflected in the skeleton motion data for conversion.

The conversion unit 4 corresponds to the type of motion of the skeleton motion data for conversion and corresponds to the expression specified by the expression specifying data, and the spatial parameters a ⁱ and b ⁱ of each segment i in the storage unit 5 and Using the time parameter m ^ _i , the skeleton motion data for conversion is converted for each segment in the principal component space.

  In the conversion unit 4, the physical quantity conversion unit 11, the principal component analysis unit 12, the segment generation unit 13, and the time axis normalization unit 14 are the same as the corresponding units 11, 12, 13, and 14 of the learning unit 3. Is omitted. Hereinafter, the space axis conversion unit 21, the time axis conversion unit 22, and the position space restoration unit 23 according to the conversion unit 4 will be described in detail.

[Space axis conversion unit]
The spatial axis conversion unit 21 corresponds to the type of motion of the skeleton motion data for conversion, and corresponds to the expression specified by the expression specifying data, and the spatial parameters a ⁱ and b of each segment i in the storage unit 5. ^{Using i} , skeleton motion data for conversion is converted for each segment in the principal component space. At this time, the space axis conversion unit 21 performs linear regression calculation according to equation (11) using the differential value of the principal component coordinates.

Here, q ⁱ (t) is a differential value of the j-th principal component coordinate in the t-th frame of the i-th segment with respect to the skeleton motion data reflecting the expression specified by the expression specifying data. k ⁱ (t) is a differential value of the j-th principal component coordinate in the t-th frame of the i-th segment regarding the skeleton motion data for conversion. a ^ ⁱ and b ^ ⁱ correspond to the type of motion of the skeleton motion data for conversion, and correspond to the expression specified by the expression specifying data, the spatial parameters a ⁱ of the segment i in the storage unit 5, b ⁱ .

The space axis conversion unit 21 performs integration on the differential values q ⁱ (t) of all segments calculated by the linear regression calculation according to the equation (11). The principal component coordinates of the first frame are used as initial values for this integration calculation. As described above, the linear regression calculation is performed using the differential value of the principal component coordinates, and then integration is performed, whereby the connection between the segments can be made smooth.

[Time axis conversion unit]
The time axis conversion unit 22 sets the time parameter m ^ _i of each segment i in the storage unit 5 corresponding to the type of motion of the skeleton motion data for conversion and corresponding to the expression specified by the expression specifying data. The integral value of the principal component coordinates calculated by the space axis conversion unit 21 is converted for each segment. At this time, the time axis conversion unit 22 converts the interpolation ratio according to equation (12).

However, t ¹ _i is the interpolation ratio of the i-th segment regarding the skeleton motion data reflecting the expression specified by the expression specifying data. t ⁰ _i is an interpolation ratio of the i-th segment regarding the skeleton motion data for conversion. m ^ ⁱ is a time parameter of the segment i in the storage unit 5 corresponding to the type of motion of the skeleton motion data for conversion and corresponding to the expression specified by the expression specifying data.

The time axis conversion unit 22 down-samples the time axis with respect to the integral value of the principal component coordinates calculated by the space axis conversion unit 21 at the interpolation ratio t ¹ _i calculated by the equation (12). The “cubic spline interpolation” technique can be used for this downsampling.

[Position space restoration section]
The position space restoration unit 23 converts the first principal component coordinates to the r-th principal component coordinates converted by the time axis conversion unit 22 into the position space according to the above equation (8).

  According to the above-described embodiment, when moving image content reflecting a specific expression is generated, an effect of improving robustness and adapting to non-periodic movement can be obtained. In addition, since the skeleton motion data reflecting the specified expression (for example, specific emotion expression) can be automatically generated with respect to the neutral skeleton motion data, the user's It can contribute to the improvement of satisfaction.

  The above is description of the conversion part 4 which concerns on this embodiment. The conversion unit 4 obtains skeleton motion data as a result in which the expression specified by the expression specifying data is reflected in the skeleton movement data for conversion. The output unit 6 outputs the skeleton motion data as the reflected result.

  As mentioned above, although embodiment of this invention was explained in full detail with reference to drawings, the specific structure is not restricted to this embodiment, The design change etc. of the range which does not deviate from the summary of this invention are included.

Further, a program for realizing the above-described moving image content generation device 1 may be recorded on a computer-readable recording medium, and the program recorded on the recording medium may be read into a computer system and executed. Here, the “computer system” may include an OS and hardware such as peripheral devices.
“Computer-readable recording medium” refers to a flexible disk, a magneto-optical disk, a ROM, a writable nonvolatile memory such as a flash memory, a portable medium such as a DVD (Digital Versatile Disk), and a built-in computer system. A storage device such as a hard disk.

Further, the “computer-readable recording medium” means a volatile memory (for example, DRAM (Dynamic DRAM) in a computer system that becomes a server or a client when a program is transmitted through a network such as the Internet or a communication line such as a telephone line. Random Access Memory)), etc., which hold programs for a certain period of time.
The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.
The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

DESCRIPTION OF SYMBOLS 1 ... Moving image content production | generation apparatus, 2 ... Input part, 3 ... Learning part, 4 ... Conversion part, 5 ... Memory | storage part, 6 ... Output part, 11 ... Physical quantity conversion part, 12 ... Principal component analysis part, 13 ... Segment generation part , 14 ... time axis normalization unit, 15 ... space axis linear regression unit, 16 ... time axis linear regression unit, 21 ... space axis conversion unit, 22 ... time axis conversion unit, 23 ... position space restoration unit

Claims (5)

For each type of motion, using the skeleton motion data of a specific representation to be learned and the neutral skeleton motion data, the neutral skeleton motion data is converted into the skeleton motion data of the specific representation in the principal component space of the skeleton motion data. A learning unit that calculates a parameter of a conversion function for conversion into
A storage unit for storing the calculated parameters for each type of movement;
An input unit for inputting neutral skeleton motion data of the type of motion to be converted;
Using the calculated parameter corresponding to the type of motion to be converted, the input neutral skeleton motion data is converted into skeleton motion data of a specified specific representation in the principal component space by the conversion function. A conversion unit for conversion;
A moving image content generation device comprising:   The moving image content generation apparatus according to claim 1, wherein the learning unit obtains a spatial parameter and a time parameter by a linear regression method for each segment in the principal component space of the skeleton motion data. The learning unit uses a differential value of principal component coordinates when calculating the spatial parameter by a linear regression method,
The conversion unit performs conversion on the differential value of principal component coordinates of the input neutral skeleton motion data by the conversion function using the spatial parameter, and integrates the converted differential value.
The moving image content generation device according to claim 2. For each type of motion, using the skeleton motion data of a specific representation to be learned and the neutral skeleton motion data, the neutral skeleton motion data is converted into the skeleton motion data of the specific representation in the principal component space of the skeleton motion data. A learning step for calculating parameters of a conversion function for conversion into
Storing the calculated parameters for each type of motion;
An input step for inputting neutral skeleton motion data of the type of motion to be converted;
Using the calculated parameter corresponding to the type of motion to be converted, the input neutral skeleton motion data is converted into skeleton motion data of a specified specific representation in the principal component space by the conversion function. A conversion step to convert;
A moving image content generation method comprising: For each type of motion, using the skeleton motion data of a specific representation to be learned and the neutral skeleton motion data, the neutral skeleton motion data is converted into the skeleton motion data of the specific representation in the principal component space of the skeleton motion data. A learning step for calculating parameters of a conversion function for conversion into
Storing the calculated parameters for each type of motion;
An input step for inputting neutral skeleton motion data of the type of motion to be converted;
Using the calculated parameter corresponding to the type of motion to be converted, the input neutral skeleton motion data is converted into skeleton motion data of a specified specific representation in the principal component space by the conversion function. A conversion step to convert;
A computer program for causing a computer to execute.

Images