JP3422684B2 - Data processing method, and storage medium storing program according to the method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data processing method capable of obtaining a free curve approximating the shape of a data string from a series of digital data strings, and a storage medium storing a program according to the method.

[0002]

2. Description of the Related Art A technique for preliminarily compressing and holding a series of digital data strings and restoring the original digital data string from the compressed data when needed is often used in various data processing. This is a required technology. For example, computer graphics (C
In G) video production, game production, CG video production of personal computer software, etc., the motion dump data expressing the movement of a certain character has a relatively large capacity, so when storing it, compress it to a storage device. The method is to store the data and restore the original motion dump data from the compressed data to display the compressed data.

Hereinafter, processing of the above-mentioned motion dump data in game production will be described as an example. 2. Description of the Related Art In recent years, a so-called motion capture system has been known in which a person actually takes in the motion dump data representing the motion of a character in a game (for example, a fighting game) by making the same motion. For example, an object that reflects light is attached to some measurement points (joints, etc.) of a human body, and the motion of that person is actually captured by a video camera, and the motion of that object is captured. The change in the three-dimensional coordinates of the measurement point and the change in the angle around each axis are captured as motion dump data. A plurality of parameters such as position coordinates and angles are taken in for each measurement point, and each parameter is, for example, 1
Since one data is fetched in a unit of / 60 seconds, the entire capacity of the motion dump data is often large even if the motion is for several seconds.

The captured motion dump data is C
The motion dump data is stored in a D-ROM or the like together with the game program, and when the character moves in accordance with the progress of the game, the motion dump data is read, and the character is displayed and moved accordingly. Since the storage capacity of a storage medium such as a CD-ROM is limited, it is desirable to compress and store the capacity of motion dump data as much as possible. Conventionally, motion dump data for each frame (one screen displayed at a predetermined interval (for example, 1/60 seconds) is called a frame) is differentially compressed and stored, and is expanded and restored when used to restore the original motion. The method of obtaining dump data was adopted.

On the other hand, at the game production site, motion
Obtained by automatically extracting some feature points (keyframes) from the motion dump data (here, paying attention to one parameter) captured by the capture system, and performing optimal spline conversion from the extracted keyframe data. CG software is used which stores a coefficient defining a free curve in a storage medium and generates and outputs a data sequence approximating the original motion dump data from the free curve when used.

[0006]

According to the method of differentially compressing and storing the above-mentioned data and decompressing and restoring the data at the time of use to obtain the original data, the original data can be accurately restored. There is an inconvenience that there is a limit to increasing the degree. Even if another compression method is used,
Basically, this is a method of compressing and retaining data so that all frames can be restored to their original state, and there is no change in the degree of compression.

Further, in the prior art, it was basically assumed that the rate at which the motion dump data was taken in was the same as the rate at which the motion dump data was displayed. For example, when each frame is captured every 1/60 seconds, the display using the data of those frames is performed every 1/60 seconds when displaying. However, in recent games and the like, various display modes are often required, and for example, a certain motion may be displayed in slow motion. In the case of slow motion, the capture rate of motion dump data and the display rate are different, and it becomes necessary to interpolate between frames. In this case, conventionally, a frame is interpolated and displayed between frames.
Therefore, there is an inconvenience that the movement becomes awkward.

On the other hand, the conventional CG software does not retain the motion dump data of all frames, but retains a free curve obtained from some feature points (key frames) in advance, and based on the free curve when used. Since a data string that approximates the original motion dump data is output, the degree of data compression can be increased. However, in the CG software currently used, the difference between the restored data and the original data is large, and there are many problems in practical use.
Hereinafter, an example of processing using conventional CG software will be described.

FIG. 9 is a graph of motion dump data showing a change (for example, a rotational movement of the clavicle of a person) of a certain parameter (positional coordinate or angle) at a certain measurement point when the person actually performs a certain action. Is. The horizontal axis indicates time, that is, a frame, and here, a part of the motion dump data of about 100 frames (around the horizontal axis of 75 to 100 frames is omitted) is shown. The vertical axis represents the data value. Figure 1
2 shows an example in which the motion dump data of FIG. 9 is processed by conventional CG software. A dotted line 1201 is a graph showing the input motion dump data. 1211 and 12
Reference numeral 12 is a key frame automatically extracted by the CG software. Here, adjustment was made so that key frames of about 20 frames were extracted from the motion dump data of about 100 frames (however, about 85 to 100 frames on the horizontal axis are omitted). At the points of each key frame, auxiliary lines 1221 and 1222 called levers are displayed.
A solid line 1202 shows a data string (free curve which is the result of the optimal spline) restored from these key frames. As can be seen from this figure, dotted line 1 showing the original data
There is a large discrepancy between 201 and the solid line 1202 indicating the restored data string, which shows that there is a practical problem. Of course, it is obvious that the original motion dump data can be restored with high accuracy by setting a large number of key frames, but in that case, the compression degree of the data cannot be increased.

In view of the above-mentioned problems in the conventional form, the present invention uses a smaller amount of data and a high precision of the original data when a free curve approximating the shape of the data sequence is generated from a series of digital data sequences. The purpose is to be able to approximate the shape. Another object is to be able to approximate the original data so that the motion does not become awkward when performing slow motion display.

[0011]

In order to achieve the above object, the invention according to claim 1 provides a predetermined program on a processing device.
By operating the ram, a series of digital data string, a data processing method for obtaining the optimized free curve approximating the shape of the data string, the program, before
From the series of digital data strings, the data value in the y-axis direction is expressed by the formula of the B-spline curve, and the x-axis direction is the simple B-spline curve expressed by the formula having a linear relationship with the parameter of the B-spline curve. the best of the table is the data structure
It is characterized by generating a free-form curve .

[0012] The invention according to claim 2, in claim 1, the formula of B- spline curve representing the data values of the y-axis direction, the B- start and end points of the control points which determine the shape of the spline curve There deformed to match the start and end points of the previous SL series of digital data sequence,
Optimized free curve of data structure expressed by the modified formula
Is generated .

According to a third aspect of the present invention, in the first aspect, with respect to the expression of the B-spline curve expressing the data value in the y-axis direction, before and after the position of the start point of the series of digital data strings. After the position of, a control point is newly set, and the data value of the control point is set so that the generated B-spline curve passes through the start point and end point of the series of digital data strings.

According to a fourth aspect of the present invention, in a data processing method for obtaining an optimized free curve that approximates the shape of a data string from a series of digital data strings, a start point, an end point, and a maximum are obtained from the series of digital data strings. A keyframe extracting step of automatically extracting points, minimum points, maximum points, minimum points, inflection points, and / or minimum points of curvature radius as keyframes, and a predetermined reduction condition from the automatically extracted keyframes A keyframe reduction step of reducing keyframes, and a free curve that approximates the original data string from the remaining keyframes after reduction, where the data values in the y-axis direction are expressed by the B-spline curve formula, and the x-axis A free curve generation for generating a free curve represented by a simple B-spline curve expressed by a formula whose direction has a linear relationship with the parameter of the B-spline curve. A step, a determination step of determining whether or not the generated free curve approximates the original data string with a predetermined degree of approximation, and an approximation of the original data string with a predetermined degree of approximation by the determination step. If judged,
Storing data defining the free-form curve.

The invention according to claim 5 relates to a storage medium storing a program according to the data processing method according to any one of claims 1 to 4.

[0016]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 shows a block configuration of a data processing apparatus according to the first embodiment of the present invention. This device is CP
U (central processing unit) 101, random access memory (RAM) 102, read only memory (ROM) 10
3, a display device 104, an input device 105, a hard disk 106, a motion capture system 107, and a bus line 108. This apparatus may be, for example, a general-purpose workstation or personal computer to which the motion capture system 107 is connected.

The CPU 101 is a processing device that controls the operation of the entire device. RAM 102 is CPU 101
Is a memory used as an area for loading programs to be executed by, and a work area for data to be processed. ROM1
A non-volatile memory 03 stores basic software such as a BIOS (Basic Input / Output System) and various parameters. The display device 104 is a CP
It is a display that displays various kinds of information in response to an instruction from U101. The input device 105 is a keyboard, a mouse or the like that is operated by a user to give an instruction to the present device. The hard disk 106 is an auxiliary storage device that stores programs and data. Motion capture
The system 107 is a system that captures motion dump data by capturing a motion of a person.

By operating the input device 105, the user can operate the motion capture system 107 to capture the motion dump data. The captured motion dump data is temporarily stored in the hard disk 106. In addition, the user operates the input device 105 to obtain a free curve that approximates the shape of the motion dump data with high accuracy, and as a result, compresses the data (described later in detail with reference to FIG. 2 and the like). ) Can be done. At the time of display, a data string that approximates the original motion dump data with a predetermined accuracy can be generated and output (restored) based on the obtained free curve. The coefficient data that defines the obtained free curve is
It is stored in the hard disk 106.

The coefficient data defining this free curve is read out, for example, a CD-ROM together with the game program.
Etc. When the game is executed by mounting the CD-ROM on a game machine, a data string is generated and output based on the free curve as needed. Since this free curve is an approximation of the original motion dump data with a predetermined accuracy, it is possible to cause the character in the game to perform a predetermined motion by using this data string.

The processing program for realizing the data processing method according to the present invention for obtaining a free curve approximating the original motion dump data and compressing the data is a CD-ROM.
It is provided in various storage media such as, and is installed in the hard disk 106 in advance. CPU 101
Executes the processing by reading this program onto the RAM 103 and executing it.

Next, the procedure of the data processing program executed by the apparatus of FIG. 1 will be described. FIG. 2 is a flowchart showing an outline of a flow of processing for determining an optimum key frame and compressing data by a free curve, which is executed by the apparatus of FIG. As a premise, it is assumed that the hard disk 106 stores the motion dump data to be processed.

FIG. 3 shows an example of motion dump data which is the processing target of the processing of FIG. The motion dump data is the data captured by the motion capture system 107, for example, the movements of several measurement points of the human body, and therefore the data for each measurement point (for example,
Data showing the movement of the shoulder, data showing the movement of the elbow, data showing the movement of the wrist, etc.). The data for one measurement point is made up of data for each of several parameters (for example, three-dimensional coordinates, angle around the coordinate axis, etc.) indicating the movement of the measurement point. The data for one parameter consists of frame data for that parameter at predetermined time intervals. The example of the motion dump data shown in FIG. 3 shows data (for example, the rotation angle of the human clavicle) about a certain parameter at a certain measurement point. The horizontal axis shows the frame number (that is, time) i, and the vertical axis shows the data value y of the parameter. Here, the number of data is n + 1 (hence i is 0 to n), and the data value of the frame number _i is represented by y _i . In the following, a process for data of a certain parameter at such a measurement point will be described, but a process for data of another parameter and a process for data of a parameter at another measurement point are also
The same is done.

In the process of FIG. 2, first, in step 201,
The motion dump data is read and key frame candidate points are extracted. For example, among the motion dump data to be processed, the start point frame and end point frame, the maximum point frame and the minimum point frame, the maximum point frame and the minimum point frame, the inflection point frame, and the minimum radius of curvature. The point frame is extracted and used as the first key frame (candidate point). After step 201, in step 202, the extracted key frames are arranged and arranged in order of frame number. Next, in step 203, the keyframes satisfying the predetermined reduction condition are reduced and the remaining keyframes are left. As a reduction condition, for example, important keyframes such as the start point, the end point, the maximum point, and the minimum point are excluded, and adjacent frames are extracted as keyframes from other keyframes, and the data of those keyframes is extracted. Conditions such as reducing one key frame when the difference in value is less than or equal to a predetermined value are used.

Next, in step 204, step 20
In the same manner as in 2, the key frames are arranged (sorted) in the order of frame numbers. In step 205, a free curve approximating the original motion dump data (FIG. 3) is generated based on the key frame data remaining at that time.
The method of generating the free curve in step 205 will be described in detail below.

In the present invention, a free-form curve is represented by using a simple B-spline curve which is a slight modification of the equation of the true B-spline curve. FIG. 4 shows an example of representation by a simple B-spline curve. x-axis is frame (time axis)
, And the y-axis shows the data value. On the data values of each frame, suitable keyframe i _0, such as _{401~403, i 1, i 2,} ..., i m, ... ( referred to as a control point to point key frames) extracted and the The simple B-spline curves 411 and 412 are expressed using the control point data values Q ₀ , Q ₁ , Q ₂ , ..., Q _im . The simple B-spline curve is expressed by (1)-
It is expressed by equation (5).

[0027]

[Equation 1]

Expressions (1) to (5) of the equation 1 are expressed by the following equation 2
~ Rewrite as in Expressions (6) to (12).

[0029]

[Equation 2]

[0030]

[Equation 3]

[0031]

[Equation 4]

From equations (6) to (12) of equations 2 to 4,
The simple B-spline curve is represented by the following equation (13).

[0033]

[Equation 5]

The original motion dump data (FIG. 3) is approximated by the equation (13), and the error of 2 is calculated by using the least square method.
Th power sum determines the data value Q _m of the control points so as to minimize. The conditional expression is given by the following Expression 6.

[0035]

[Equation 6]

It should be noted that the weighting factor w _i is introduced here, and the key frames considered to be significant are processed with their weights higher than those of other frames, and the approximation formula given by y (x) is processed as much as possible. It passes near the data value of the frame. The weighting factor w _i may be determined based on the significance of each frame (how significant it is to approximate the shape of the original motion dump data). For example, the start point, the end point, The maximum and minimum points are w _i = 10, the maximum points (excluding the maximum points) and the minimum points (excluding the minimum points) are w _i = 5, and the inflection points are w _i = 2, and other keys The frame may be appropriately determined such that w _i = 1.

The equation (6) is a variational problem, and its stopping principle is given by the following equation (7). That is, Q _m that satisfies the conditional expression of Expression 6 can be obtained by solving Expression 7.

[0038]

[Equation 7]

From the equation 7, the following equation 8 can be derived.

[0040]

[Equation 8]

The following Equation 9 can be obtained by rearranging Equation 8.

[0042]

[Equation 9]

The equation (9) is given by the equation (18) of the following equation (10).

[0044]

[Equation 10]

This equation (19) is expressed by Q _m (m = 0, 1,
..., because it is n _m based on simultaneous linear equations for n _m -1), it is possible to obtain the Q _m by solving this.
Once Q _m (m = 0, 1, ..., N _m −1) is determined, the original motion dump data (FIG. 3) is approximated by substituting the Q _m into Equation (13). The free curve (simple B-spline curve) to be obtained is obtained. As described above, the generation of the free curve in step 205 of FIG. 2 is executed.

After obtaining the free curve in step 205, optimization is determined in step 206. The optimization determination is performed by, for example, calculating the root mean square error or the maximum deviation degree between the obtained free-form curve and the original motion dump data, and whether those values are within a predetermined value (that is, a predetermined approximation degree). And whether or not a free curve approximating the original motion dump data was obtained in.). If it is determined in step 206 that further optimization is necessary, the process returns to step 203 again, and the process is repeated from the process of removing the key frame under another reduction condition. Steps 203-2
The processing of 06 is the same as that described above. Step 206
If it is determined that the optimization has been performed in step 10, the data value Q _m of the control point at that time is set to the hard disk 10
Store in 6. These data are stored in a CD-ROM or the like together with the game program, and display data of each frame is generated and displayed on the basis of the free curve thereof as needed during the game, so that the data can be displayed to the character in the game. Can be operated.

The simple B-spline curve is obtained by modifying the equation of the true B-spline curve into a simple form. The equation of the true B-spline curve is shown in Equation 11.

[0048]

[Equation 11]

As can be seen by comparing the equation (3) of the equation 1 representing the simplified B-spline curve and the equation of the above equation 11, in the true B-spline, the data value x in the x-axis direction is obtained.
While (η) is expressed by a cubic expression of the parameter η, the data value x is linearly related to the parameter η in the simplified B-spline curve. As a result, the data value x and the parameter η have a one-to-one correspondence, and the calculation can be performed easily and quickly. Also, the data value y in the y-axis direction
For (η), the equation of the intrinsic B-spline is used as it is for the simple B-spline.

In the conventional CG tool, the optimum spline compression based on the cubic Hermitian curve is mainly used.
As compressed data obtained by this technique, three data of the frame number for each key frame selected for the optimal spline, the value of the data, and the slope of the data become one set, and the number of selected key frames Only needed. On the other hand, according to the B-spline curve used in the above-described embodiment of the present invention, two sets of data, that is, the frame number and the data value for each key frame selected for the optimal spline are set. , You need as many keyframes as you have selected. Therefore, the compression ratio can be improved as compared with the data compression by the third-order Hermitian curve.

FIG. 7 shows an example of approximating motion dump data (indicated by black dots) having a shape of a sine wave, using the technique of the above-described embodiment of the present invention. 701-706
The points indicated by are the control points at the positions of the key frames, and 711 to 713 are straight lines connecting these control points. 720 shows a simplified B-spline curve generated from these control points. It can be seen that the original data can be approximated with high accuracy with few control points. In the method of the above embodiment, it is uncertain whether the generated simple B-spline curve passes through the start point 721 and the end point 722 of the original motion dump data. Further, since the control points 701 and 706 at the start point and the end point do not coincide with the start point and the end point, the data values at the start point and the end point need to be calculated.

Next, a first modification of the above embodiment will be described. In the method of the above embodiment, the generated simple B-spline curve does not always pass through the start point and the end point of the motion dump data. However, since the data values of the start point and the end point of the motion dump data are important and significant in terms of motion, the generated simple B
-I want to make sure that the spline curve passes through the start and end points of the motion dump data. Therefore, in the first modified example, both ends of the motion dump data, that is, before the control point Q _{0 at} the start keyframe position and at the end point of the control point Q _0.
A control point was added after _nm-1 to ensure that the generated simple B-spline curve would pass the start and end points of the motion dump data.

Specifically, the equations (1) and (5) of the above-mentioned embodiment are replaced by the equations (1) 'and (5)' of the following equation 12. The difference between the equations (1) and (5) of the equation 1 and the equations (1) ′ and (5) ′ of the equation 12 is that the variables of Q ₋₁ and Q _nm are added.

[0054]

[Equation 12]

Equations (2) to (4) may be the same as equation (1). Further, Q _{0 to} Q _{m to} Q _nm-1 can be obtained by the method of the above embodiment.

Equation (1) 'is the starting point (x = i ₀ = 0, y
In order to pass (i ₀ ) = y ₀ ), the value of Q ₋₁ may be determined so as to satisfy the following expression 13.

[0057]

[Equation 13]

Similarly, the expression (5) 'is the end point (x = i _nm-1 =
In order to pass 0, y (i _nm-1 ) = y _nm-1 ) without fail, the value of Q _nm should be determined so as to satisfy the following equation (14).

[0059]

[Equation 14]

From the above, as the new data, the above-mentioned Q
By adding _-1 and Q _nm , the simple B-spline curve can be sure to pass through the start point and the end point of the motion dump data.

Next, a second modification of the above embodiment will be described. According to the first modification, the simple B-spline curve can pass through the start point and the end point of the motion dump data without fail. However, in the first modified example, new data must be added, and from the viewpoint of increasing the compression rate, such data should not be added if possible. Further, although it is guaranteed that the generated simple B-spline curve passes through the start point and the end point of the motion dump data, the data values at the start point and the end point must be calculated. On the other hand, in the production of game images, the data values of the start point and the end point are often used directly in motion interpolation calculations, etc., and it is troublesome to calculate the data value of the start point and the end point from the data value of the control point when necessary. Is.

Therefore, in the second modified example, the simple B-spline equation used in the above embodiment is modified so that the start point and the end point of the motion dump data and the control point always match. Specifically, the above number 1
Expressions (1) to (5) are expressed by, for example, the following Expression 15 (1) ″
~ (5) ".

[0063]

[Equation 15]

The equation (1) "of the equation 15 replaces Q ₀ of the equation (1) of the equation 1 with the data value y _{0 of the} starting point, and further, the portion 3η ³ − of the coefficient of the second item is enclosed. 6η ² +4 to f1
(Η) = - η + 1 is added, parts are enclosed in braces coefficients of the third item ^{-3 (η-1) 3 -6} (η-1) 2 +4 to g1 of (η) = η-1 This is the added formula. The expression (2) "of the expression 15 is an expression in which Q ₀ of the expression (2) of the expression 1 is replaced with the data value y _{0 of the} starting point. The expression (3)" of the expression 15 is the expression (3) of the expression 1. It is the same formula as the formula. The formula (4) ”of the equation 15 is
This is an expression in which Q _{nm-1 in the} expression (4) of Expression ₁ is replaced with the end point data value y _n-1 . The formula (5) ”of the formula 15 is the formula (5) of the formula 1.
Q _{nm-1 in the} equation is replaced with the data value y _{n -1} at the end point, and the part of the coefficient of the second item enclosed in parentheses ³ η ³ −6 η ²
Addition of f2 (η) =-η to +4, and the part enclosed in braces of the coefficient of the third item -3 (η-1) ³ -6 (η-1) ²
This is an expression obtained by adding g2 (η) = η to +4.

A modification of the above equation is that when η = 0, (1) "
The expression always becomes the data value y _{0 at the} start point, and when η = i _nm−1 , the expression (5) ″ always becomes the data value y _{n−1 at the} end point. Further, in the B-spline, the affine In order to maintain the property, the sum of the coefficient of each term must be 1. However, it is modified so that the condition is satisfied. Therefore, the data (start point and end point) for motion interpolation calculation, which is essential in game video production, can be directly used.

It should be noted that, since the equations (1) "to (5)" in the above equation 15 are modified, the equations (6) to in the equations 2 to 4 are changed.
Expression (12) is expressed by the following Expressions 16 to 17 (6) ″ to (1
2) ", and thereafter, the process after the formula (13) may be performed in the same manner.

[0067]

[Equation 16]

[0068]

[Equation 17]

FIG. 8 shows an example in which motion dump data (indicated by black dots) similar to the shape of a sine wave similar to that in FIG. 7 is approximated by the method of the second modification. 801 to 806
The points indicated by are control points at the positions of the key frames, and 811 to 813 are straight lines connecting these control points. 820 shows the simple B-spline curve generated from these control points. It can be seen that the original data can be approximated with high accuracy with a small number of control points, and the start point and end point are treated as control points, and at the same time, a curve is generated so as to pass through the start point and end point.

Further, FIG. 10 is a graph in which the control points obtained by the method of the second modification described above are connected by a straight line for the motion dump data of FIG. 9 described in the section of the problem to be solved by the invention. Indicates. The black dots are control points. FIG. 11 shows a simple B-spline curve generated from the control points. As can be seen from the figure, a curve that approximates the original motion dump data with high accuracy can be obtained with few control points.

Next, a third modification of the above embodiment will be described. In the method of the above-mentioned embodiment, the key frame has a start point, an end point, a maximum point, a minimum point, a maximum point, a minimum point,
The inflection point and the minimum point of the radius of curvature are extracted as a key frame. However, the distribution may be biased, such as when there are many keyframes and when there are few keyframes. In this case, I want to delete unnecessary keyframes in many parts and allocate keyframes in few parts.

Therefore, in the third modification, the motion dump data is mechanically divided and the simple B-spline curve is generated in each section by the method of the above embodiment. Then, in each section, the root mean square error is calculated between the generated simple B-spline curve and the original motion dump data, and the key frame of the section having the smallest root mean square error (the key frame of the section Among them, select the keyframe with the smallest error between the generated simple B-spline curve and the original motion dump data, and apply it to the section where the root mean square error is the largest (the data error in that section is the largest). Put the keyframe in the place). A key frame is newly added to a section other than the section having the smallest mean squared error and having a mean squared error larger than the specified value of the mean squared error (that section Put the keyframe at the point where the data error is the largest). As described above, it is possible to eliminate the biased distribution of the key frames.

As a method of eliminating the biased distribution of the key frames, the following fourth modification may be adopted. Fourth
In the modified example, the key frame is placed in a flat section more intensively than a section in which the motion dump data greatly changes.

Specifically, in step 203 of FIG.
The keyframe reduction process is as follows. First
All keyframe candidate points (start point, end point, maximum point, maximum point
Initial values for small points, local maximum points, local minimum points, inflection points, etc.)
Flag value flag [m] (where m is the frame number after sorting)
And m = 0, 1, ..., N_m-1) is prepared. Next
Set the flag value flag [m] of the important keyframe to 4
It Here, the flag values of the start point and the end point are set to 4.
Next, the height difference between all adjacent keyframe candidates is
If there are more than one, a process of adding 1 to each flag value is performed. You
Nachi, ｜ y_im-Y _{im + 1}│ ≧ α (y_max-Y_min) Is
It is determined whether or not it stands, and if it holds, flag [m] and fal
Increment g [m + 1].

5 (a) to 5 (h) show patterns of flag values, respectively. Black dots indicate keyframe points. The positional relationship between the upper and lower black circles indicates the relative size of the keyframe data values (the data values at the upper points are large and the data values at the lower points are small). "Large" attached to the black circle and the line segment connecting the black circles means that these points are adjacent keyframes that satisfy the above condition | y _im −y _{im + 1} | ≧ α (y _max −y _min ). Indicates. “Small” indicates that the key frames are adjacent key frames that do not satisfy the above conditions. The numbers "1" and "2" written near the black circles indicate the flag value flag [m] of the key frame.

Further, the average point slope of the motion dump data in the vicinity of a certain key frame candidate point is obtained, and if it is below the condition, 1 is added to the flag value. FIG. 6 shows such a key frame candidate point 601. Here, when the approximate linear expression is y = a ₀ + a ₁ · x, the least squares method gives
a ₀ and a ₁ are given by the following _equation 18. As described above, the flag value near the center of the flat section is increased, and then the key frame candidates having the flag value of 1 or less are reduced.

[0077]

[Equation 18]

As a result, the key frames 601 and the like in FIG. 6, which are reduced in the above-described embodiment, are not reduced, so that the uneven distribution of the key frames can be eliminated.

[0079]

As described above, according to the present invention,
Since a simple B-spline curve is used when a free curve that approximates the shape of a data string is generated from a series of digital data strings, the shape of the original data can be approximated with less data with high accuracy. it can. Also, simple B
-Because the spline curve can always pass through the start point and end point of the motion dump data, it is possible to join continuously when joining multiple motion spline data. In addition, the start point and the end point can be treated as control points, and at the same time, a curve can be generated so as to pass through the start point and the end point without fail, so data (start point and end point) for motion interpolation calculation, which is essential in game video production, can be directly used. it can. Furthermore, it can be used to generate smooth continuous motion by smooth interpolation of different movements of a character. Smooth motion can be realized even when the motion of the character is played back in slow motion.

[Brief description of drawings]

FIG. 1 is a block configuration diagram of a data processing device according to an embodiment of the present invention.

FIG. 2 is a flowchart showing an outline of a flow of data compression processing using a free curve.

FIG. 3 is a diagram showing an example of motion dump data to be processed.

FIG. 4 is a diagram showing a representation example of a simple B-spline curve.

FIG. 5 is a diagram showing a pattern of flag values.

FIG. 6 is a diagram showing an example in which a key frame is provided near the center of a flat portion.

FIG. 7 is a result of simple B-spline curve interpolation (part 1).
Showing

FIG. 8 is a result of simple B-spline curve interpolation (part 2).
And a diagram showing an example in which the start point and the end point are always passed

FIG. 9 is a diagram showing an example of motion dump data.

FIG. 10 is a diagram showing a graph in which control points are connected by straight lines.

FIG. 11 is a diagram showing an example of a result of simple B-spline curve interpolation.

FIG. 12 is a diagram showing an example of cubic Hermitian curve interpolation by conventional CG software.

[Explanation of Codes] 101 ... CPU (Central Processing Unit), 102 ... Random Access Memory (RAM), 103 ... Read Only Memory (ROM), 104 ... Display Device, 105 ... Input Device, 1
06 ... Hard disk, 107 ... Motion capture system.

─────────────────────────────────────────────────── ─── Continuation of front page (56) Reference JP-A-3-127186 (JP, A) JP-A-6-215094 (JP, A) JP-A-6-318258 (JP, A) JP-A-3- 160579 (JP, A) JP-A-4-71078 (JP, A) Akira Sakurai, 3 persons, "Introduction to Spline Function", Tokyo Denki University Press, June 30, 1981, 1st edition, p. 4 (58) Fields investigated (Int.Cl. ⁷ , DB name) G06T 13/00 G06F 17/17 G06T 11/20 CSDB (Japan Patent Office)

Claims (5)

(57) [Claims] 1. A data processing method for obtaining an optimized free curve approximating the shape of a data sequence from a series of digital data sequences by operating a predetermined program on a processing device, wherein the program comprises: From the series of digital data strings, a data value in the y-axis direction is expressed by a B-spline curve formula, and an x-axis direction is expressed by a simple B-spline curve expressed by a formula having a linear relationship with the parameters of the B-spline curve. A data processing method, characterized in that an optimized free curve of a represented data structure is generated. 2. The B-spline curve expression expressing the data value in the y-axis direction is defined by the starting point and the ending point of a control point that determines the shape of the B-spline curve. The data processing method according to claim 1, wherein the optimized free curve of the data structure represented by the modified equation is generated by transforming the optimized free curve. 3. A control point is newly set before the position of the start point and after the position of the end point of the series of digital data strings with respect to the B-spline curve expression expressing the data value in the y-axis direction. The data processing method according to claim 1, wherein the data value of the control point is set so that the generated B-spline curve passes through the start point and the end point of the series of digital data strings. 4. A data processing method for obtaining an optimized free curve approximating the shape of a data string from a series of digital data strings, wherein a starting point, an end point, a maximum point, a minimum point, and a maximum are selected from the series of digital data strings. A keyframe extracting step of automatically extracting a point, a minimum point, an inflection point, and / or a minimum point of a radius of curvature as a keyframe; and a key for reducing keyframes satisfying a predetermined reduction condition from the automatically extracted keyframe. A frame reduction step and a free curve that approximates the original data string from the remaining key frames after reduction, where the data value in the y-axis direction is expressed by the formula of the B-spline curve, and the x-axis direction is the B-spline curve. A free curve generating step of generating a free curve represented by a simple B-spline curve expressed by an equation having a linear relationship with the parameter of And a determination step of determining whether or not the free-form curve approximates the original data string with a predetermined degree of approximation, and when it is determined by the determination step that the original data string is approximated with a predetermined degree of approximation And a step of storing data defining the free curve, the data processing method. 5. A storage medium on which a program according to any one of claims 1 to 4 is stored.