WO2012167475A1

WO2012167475A1 - Method and device for generating body animation

Info

Publication number: WO2012167475A1
Application number: PCT/CN2011/077083
Authority: WO
Inventors: 董兰芳; 陈家辉; 李德旭
Original assignee: 华为技术有限公司
Priority date: 2011-07-12
Filing date: 2011-07-12
Publication date: 2012-12-13
Also published as: CN103052973A; CN103052973B

Abstract

A method and device for generating body animation is provided, which relates to the animation technology field. The method comprises: acquiring the initial positions of the feature points of body in the image; reading the action sequences of said feature points and calculating the positions of said feature points in the current frame according to the action sequences; according to the initial positions of said feature points and the positions of said feature points in the current frame, the image of the body region in said image is deformed; the deformed image of the body region is covered on the background image in said image in order to generate animation. Animation is generated using two-dimensional image deformation technology without creating three-dimensional model, thus reducing the workload; In addition, since single image can be driven to form body animation of any action by modifying the action sequences, no analysis nor clustering on large number of images is needed, thus the method is easy to apply and the calculation amount is small.

Description

Method and device for generating body animation

The present invention relates to the field of animation technology, and in particular, to a method and apparatus for generating a body animation. Background technique

Image form animation technology refers to the technique of processing a number of images containing human body by a computer to generate human body animation. At present, image body animation technology mainly has three-dimensional model-based technology and image series-based technology.

The technology based on the three-dimensional model: First, the human body in the input image is mapped to obtain a three-dimensional human body model similar to the human body, and then the animation is generated by driving the three-dimensional human body model to perform a specified action. The advantage of this method is that once a 3D human body model is obtained, an animation of any action can be generated.

Based on the image series technology: First, several motion images of the human body are obtained from a plurality of input images, and then an image gradient is performed between the motion images to obtain an animation effect. The advantage of this method is that the amount of calculation is small and the animation speed is fast.

In the process of implementing the present invention, the inventors have found that the prior art has at least the following problems:

For 3D model-based techniques, the process of mapping the human body in the image into a 3D human body model is cumbersome and labor intensive; for the image series based technology, it is obtained by image grading between the actions of several images in the input. Animation, so you can only produce animations that already have motion in the image. Summary of the invention

In order to reduce the workload of generating a form animation, and to generate a body animation of various actions based on a single image, an embodiment of the present invention provides a method and apparatus for generating a body animation. The technical solution is as follows:

In one aspect, a method of generating a form animation is provided, the method comprising:

Obtaining an initial position of a feature point of the shape in the image;

Reading an action sequence of the feature point, and calculating a position of the feature point in the current frame according to the action sequence; and the image according to an initial position of the feature point and a position of the feature point in a current frame The image of the medium body region is image deformed;

An image of the deformed body region is overlaid on the background image in the image to generate an animation.

In another aspect, an apparatus for generating a form animation is provided, the apparatus comprising: An obtaining module, configured to acquire an initial position of a feature point of a shape in the image;

a calculation module, configured to read an action sequence of the feature point, and calculate a location of the feature point in the current frame according to the action sequence;

a deformation module, configured to perform image deformation on an image of the shape region in the image according to an initial position of the feature point acquired by the acquisition module and a position of the feature point in the current frame calculated by the calculation module;

And a generating module, configured to overlay an image of the deformed shape region of the deforming module on the background image in the image to generate an animation.

The beneficial effects brought by the technical solutions provided by the embodiments of the present invention are:

By reading the action sequence of the feature point, the position of the current frame feature point is calculated according to the motion sequence, and then the image is deformed based on the initial position of the acquired feature point, and the position of the current frame feature point is reached, thereby generating a two-dimensional image. Body animation, because the method uses two-dimensional image deformation technology, there is no need to establish a three-dimensional model, thus reducing the workload; the method drives the body to generate motion through the action sequence, and only needs to modify the motion sequence to drive a single image to form an arbitrary The shape animation of the motion does not need to analyze and cluster the body motions in a large number of images to acquire different types of motions as in the prior art, so the method is simple to implement and the calculation amount is small. DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly described. It is obvious that the drawings in the following description are only some embodiments of the present invention. Other drawings may also be obtained from those of ordinary skill in the art in view of the drawings.

1 is a flowchart of a method for generating a body animation according to Embodiment 1 of the present invention;

2 is a flow chart of a method for modeling shape parameters according to Embodiment 2 of the present invention;

3 is a schematic diagram of an upper body image of a human body according to a second embodiment of the present invention;

4 is a schematic diagram of feature points of a body according to a second embodiment of the present invention;

FIG. 5 is a schematic diagram showing the division of a body region according to Embodiment 2 of the present invention; FIG.

6a is a schematic diagram of a first image restoration provided by Embodiment 2 of the present invention;

FIG. 6b is a schematic diagram of a second image restoration according to Embodiment 2 of the present invention; FIG.

6c is a schematic diagram of a third image restoration provided by Embodiment 2 of the present invention;

7 is a flowchart of a method for pre-defining an action sequence according to Embodiment 2 of the present invention;

FIG. 8 is a flowchart of a method for generating a body animation according to Embodiment 2 of the present invention; FIG.

9 is a schematic diagram of forming an auxiliary feature line according to Embodiment 2 of the present invention; FIG. 10 is a schematic structural diagram of an apparatus for generating a body animation according to Embodiment 3 of the present invention; FIG.

FIG. 11 is a schematic structural diagram of another apparatus for generating a body animation according to Embodiment 3 of the present invention; FIG.

FIG. 12 is a schematic structural diagram of another apparatus for generating a body animation according to Embodiment 3 of the present invention; FIG.

FIG. 13 is a schematic structural diagram of another apparatus for generating a body animation according to Embodiment 3 of the present invention. detailed description

The embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

Embodiment 1

Embodiments of the present invention provide a method for generating a body animation, which can generate an animation of a corresponding body for any given image containing a human body, an animal, or a cartoon image. Referring to Figure 1, the method flow includes:

101: Obtain an initial position of a feature point of a shape in the image;

102: Read an action sequence of the feature point, and calculate a position of the feature point in the current frame according to the action sequence;

103: deforming an image of the body region in the image according to an initial position of the feature point and a position of the feature point in the current frame;

104: An image of the deformed body region is overlaid on the background image in the image to generate an animation.

The method provided by the embodiment of the present invention calculates the position of the feature point of the shape in the current frame according to the action sequence by reading the action sequence of the feature point, and then performs image deformation on the basis of the initial position of the acquired feature point to reach the current The position of the frame feature points, thus generating a two-dimensional shape animation, because the method uses a two-dimensional image deformation technique, and does not need to establish a three-dimensional model, thereby reducing the workload; the method generates motion by driving the body through the action sequence, and only needs Modifying the sequence of motions can drive a single image to form an animation of an arbitrary motion, without the need to analyze and cluster the body motions in a large number of images to obtain different types of motions as in the prior art, so the method is simple to implement, and the calculation is simple. Small amount. Embodiment 2

Embodiments of the present invention provide a method for generating a form animation, which can generate an animation of a form in an image for any given image, wherein the shape in the image can be in a human body, an animal, or a cartoon image. One kind. In this embodiment, an upper body animation of a human body, an animal, a cartoon image, or the like can be generated, and a whole body animation can be generated. In the embodiment of the present invention, an animation is generated only for an upper body image of a human body, but the present invention is not limited thereto.

The method provided by the embodiment of the present invention, after the user or the system inputs the original input image, first performs shape parameter modeling on the image, and then generates 2D (Two Dimensions, 2D) animation according to the built model, see FIG. 2, The method flow of shape parameter modeling includes:

201: Scan the original input image to obtain a feature point and a body region predefined for the shape in the image; assume that the original input image input by the user is the upper body image of the human body, as shown in FIG. 3, in order to make subsequent annotations in the image More clearly, Figure 3 shows only the outline of the upper body image of the human body. The image details in the middle of the outline are not displayed, but the actual application should be a real upper body image.

Specifically, the original input image is scanned, and the feature of the image in the image is analyzed to obtain a feature point and a body region predefined for the shape.

202: Display a predefined feature point and a body area into the image;

Specifically, the predefined feature points and the body region are zoomed and displayed to a suitable position in the image, but sometimes the feature points and the body regions displayed in the image cannot completely correspond to the corresponding positions of the body due to the inaccuracy of the positioning. Therefore, the user is required to perform precise positioning according to the displayed feature points and the body regions, as described in step 203.

203: Save the position where the predefined feature points and the shape regions displayed by the user are dragged to the corresponding positions on the shape in the image to be accurately positioned;

When positioning, the user can drag the feature points and the body area one by one to the appropriate position on the body, and also correct the fold line of the body area, which is easy to operate. After the user completes the revision, the feature points of the user positioning and the position of the body region are saved.

In the embodiment of the present invention, there are 14 feature points, as shown in FIG. Feature points can be increased or decreased. The more feature points generated, the more detailed the animation effect, and the fewer the feature points, the faster the animation will be generated.

In the embodiment of the present invention, the division of the body region is as shown in FIG. 5, and is divided into three parts: a body region (region 0), a left-hand region (region 2), and a right-hand region (region 1). The body area can also be increased, such as separating the upper arm from the lower arm, so that the image of the upper arm and the lower arm can be processed. The purpose of selecting the three regions by using the fold lines is to separate the body from the background while dividing the body into three parts for image deformation.

204: Perform image restoration on the image according to the saved position of the body region accurately positioned by the user, and obtain an image and a background image of each body region.

In the image restoration, a fast image restoration algorithm based on image average gray value commonly used in the prior art can be used, but is not limited thereto.

Specifically, when image restoration is performed, several shape regions are divided in the image, and image restoration is performed several times, and each time for different body regions is repaired. The reason why the repair is repeated is because the arm area of the shape in the image may block the body area, or the parts of the body block the background image, and the multiple repairs can restore the image information of each layer, so that the image is performed. When deformed, the arms, body and background are separated and do not affect each other.

For example, when performing image restoration on FIG. 5 in which the shape region is divided, it is necessary to repair three times: First, image repair is performed on the region 2 in FIG. 5, and FIG. 6a is obtained, and the background region blocked by the left-hand region is restored in FIG. 6a and Body area Information; then image repair is performed on the area 1 in Fig. 6a, and Fig. 6b is obtained. In Fig. 6b, the information of the background area and the body area blocked by the left-hand area and the right-hand area is restored; finally, the image of the area 0 in Fig. 6b is performed. Repair, Figure 6c, Figure 6c restores the information of the background area blocked by the left hand area, the right hand area and the body area, so that Figure 6c is the final background image, each frame animation is in the background of Figure 6c.

Further, after the modeling of the original input image is completed, the 2D animation can be generated. In the embodiment of the present invention, the action sequence of the feature points read in the 2D animation is predefined, see Figure 7. The method of predefining the sequence of actions includes:

701: predefined action basic element;

The action basic element is used to indicate the change of the position of the feature point and the number of frames of the change. The method provided by the embodiment of the present invention predefines four action basic elements, including a static basic element, a translation basic element, a rotation basic element, and a convergence basic element. .

Static basic element: Indicates that the position of the corresponding feature point is unchanged and there are no parameters.

Panning basic element: Indicates that the corresponding feature point is to shift a certain displacement within several frames. There are 3 parameters, including the two-dimensional displacement X and y of the movement of the feature point, and the number of frames that continue to move. There are many movements related to translation, such as shrugging, which is to shift the feature points 6 and 7 in Figure 4 up by a certain displacement. At the same time, the translation can also achieve the effect of the exaggerated action in the two-dimensional animation. For example, if the feature points 6 and 7 in FIG. 4 are respectively moved outward, there is an effect that the arms suddenly become strong. The unit of panning is calculated according to the default image size of 1000*1000. For the images of different sizes input, the coordinate conversion will be performed according to the size of the default image and the actual size of the input image, that is, if the size of the input image is iWidth*iHeight, Then the amount of translation x*=iWidth/1000,

The data structure of the panning primitive can be designed as follows:

Typedef struct STR— TRANS 〃 translation basic element

{

Bool bTrans; 〃 is translation

Int iX ; //translation x

Int iY ; 〃 translation y

Int iTime ; 〃 translation continuous frame number

} STR- TRANS;

Rotating elementary element: Indicates that the corresponding feature point is to be rotated around a feature point within a number of frames. There are 4 parameters, including the surrounding feature point, the rotation angle on the 2D plane of the image, and the vertical The angle of rotation in the dimension plane and the number of frames that the rotation continues. The surrounded feature points are also referred to as rotation base points, and are represented by the number of the feature points. Most of the hand movements in an animation are related to rotation, which is also consistent with the bone movements of the body. The data structure of the rotating primitive can be designed as follows:

Typedef struct STR— ROTATE 〃 rotation primitive Bool bRotate ; // whether to rotate

Int iO ; 〃 rotation base point 0

Float fTheta ; 旋转 the angle of rotation on a two-dimensional plane

Float fZ ; 〃 vertical rotation angle in the direction of the two-dimensional plane

Int iTime ; 〃 rotates the number of frames

} STR— ROTATE;

Convergence Element: Indicates that the corresponding feature point is to be moved to a feature point within several frames. There are 3 parameters, including the target feature point, the movement ratio, and the number of frames that are continuously moved. The movement ratio is the percentage of the distance from the feature point to the target feature point. If the movement ratio is 1, the feature point is moved to the position of the target feature point. If the movement ratio is 0.5, then the movement is The position to the middle of the feature point and the target feature point. The rendezvous operation is used to compensate for the defect that the position of the imaginary position cannot be reached due to the definition of the precise displacement value in the translation operation, such as placing the left hand on the right shoulder, because the coordinates of the right shoulder are not known in advance, and the translation cannot be accurately reached by the translation. . The data structure of the meeting element can be designed as follows:

Typedef struct STR—MIX 〃 convergence basic element bool bMix ;

Int iO ; 〃 target feature point

Float fRate ; 〃 moving ratio

Int iTime ; 〃 convergence continuous frame number

} STR- MIX ;

702: pre-defining an action according to a feature element of the action element and a feature point of the shape;

Specifically, each feature point corresponds to one action basic element in one action. Therefore, a specific step of predefining the action according to the motion element and the feature point of the shape is to confirm the position change of the feature point of the shape one by one in the action. When it is confirmed that the position of the feature point is unchanged, the position change of the feature point is represented by a stationary basic element;

When it is confirmed that the feature point is to be translated, the position change of the feature point is represented by a translation basic element;

When it is confirmed that the feature point is to be rotated around other feature points, the position change of the feature point is expressed by the rotation basic element;

When it is confirmed that the feature point is to move to other feature points, the position change of the feature point is represented by the convergence basic element. The following uses the shape in Figure 4 as an example. The data structure used to define the action can be designed as follows: typedef struct STR- ACTION 〃 action Bool bStill[14]; 〃Features do not move

STR— TRANS strTrans[14]; 〃 feature point translation

STR— ROTATE strRotate[14]

STR— MIX strMix [14]; 〃 feature points meet

帧The number of frames that the action continues

}STR_ACTI0N;

Among them, [14] indicates that the corresponding array has 14 array elements, because there are 14 feature points in Figure 4. It is confirmed by which action element is used to confirm the position change of the feature point in the action, and the change and parameter of the feature point position are recorded in the array element corresponding to the feature point in the corresponding action element.

The number of frames in which the motion continues is specifically to find the maximum number of consecutive frames in which the position changes in each feature point, and the maximum number of consecutive frames is used as the number of frames in which the motion continues.

Further, after defining the data structure of the action, the content of the file representing the action can be obtained, assuming that the static element is represented by s, the basic element is represented by t, the basic element is represented by r, and the basic element is represented by m. The actions of the three shapes in Figure 4 are listed below:

Ssssssssssr 8 90 60 10 r 9 -90 60 10 ss 〃 action 1 ssssssssssssm l3 0.5 8 m l2 0.5 8 //action 2 sssssssssst 0 -20 5 t 0 -20 5 t 0 -20 5 t 0 -20 5 〃 action 3 In action 1, the first ten s indicate that the feature points 0 to 9 do not move, and "r 8 90 60 10" indicates that the feature point 10 is rotated by 90 degrees around the feature point 8 on the two-dimensional plane of the image within 10 frames. Rotating 60 degrees vertically in the direction of the two-dimensional plane, "r 9 -90 60 10" indicates that the feature point 11 is rotated within a frame of 10 degrees on the two-dimensional plane of the image, and the two-dimensional plane is perpendicular to the two-dimensional plane. The direction is rotated by 60 degrees, and the last two s indicate that the feature points 12 and 13 are not moving. Action 1 The actual animation effect is to swing the left and right hands to the chest within 10 frames.

In action 2, the first twelve s indicate that the feature points 0 to 11 are not moving, and "m 13 0.5 8" indicates that the feature point 12 moves within the eight frames to the midpoint position of the line connecting the feature point 12 and the feature point 13, " m 12 0.5 8" indicates that the feature point 13 moves to the midpoint position of the line connecting the feature point 13 and the feature point 12 within 8 frames. The actual animation effect of action 2 is to make the fingertips of both hands contact within 8 frames.

In action 3, the first ten s indicate that the feature points 0 to 9 are not moving, and the last four "t 0 -20 5" indicate that the feature points 10 to 13 are shifted upward by 20 in 5 frames. Action 3 The actual animation effect is to move the left and right hands up.

This step can predefine multiple actions for the input image.

703: A sequence of actions that combine predefined actions into feature points. Specifically, by combining a plurality of predefined actions, a sequence of actions can be obtained. For example, reading an action sequence of the body in FIG. 4 is as follows:

#actl#5

s s s s s s s s s r 8 90 60 10 r 9 -90 60 10 s s

s s s s s s s s s s m l3 0.5 8 m l2 0.5 8

s s s s s s s s s t 0 -20 5 t 0 -20 5 t 0 -20 5 t 0 -20 5

s s s s s s s s s t 0 20 5 t 0 20 5 t 0 20 5 t 0 20 5

s s s s s s s s s t 0 -20 5 t 0 -20 5 t 0 -20 5 t 0 -20 5

The name of the action sequence is actl, which is decomposed into five actions.

This step can pre-combine multiple motion sequences for the input image.

Further, after the shape parameter modeling and the predefined action sequence are completed, referring to FIG. 8, the method for generating the body animation includes:

801: Read an action sequence of feature points;

Specifically, the sequence of actions of the animation to be generated is read into the memory, and the sequence of actions can be divided into several actions. This step 801 can be performed before the step 802 or before the step 803, which is not specifically limited in the embodiment of the present invention.

802: Obtain an initial position of a feature point of a shape in the image;

Specifically, the initial position of the feature points of the shape in the image is obtained in the following two ways:

Obtaining the position of the feature point of the shape in the original input image, and taking the position of the acquired feature point as the initial position of the feature point; or

The position of the feature point of the shape in the image saved at the completion of the previous motion is obtained, and the position of the acquired feature point is taken as the initial position of the feature point.

It can be seen from the above two ways that the initial position of the feature point can have two definitions, one is the position of the feature point before the image is not subjected to any animation, that is, the original input image of the user positioning when the initial shape parameter is modeled. The position of the feature point in the middle; the other is the position of the feature point of the shape in the saved image when the last complete action in the action sequence is completed.

In the method provided by the embodiment of the present invention, if the initial position of the former feature point is adopted, an animation similar to the broadcast gymnastics form is generated in the subsequent generation of the animation, that is, each action in the action sequence is based on the shape. The initial feature point position is deformed; if the initial position of the latter feature point is used, the animation of the coherent action is generated in the subsequent animation generation, that is, each action in the action sequence is completed according to the previous action. The position of the point is deformed. The method provided by the embodiment of the present invention does not limit the initial position of which feature point is adopted, and may adopt any one of the above two methods to obtain the initial position of the feature point.

803: Calculate a position of a feature point in the current frame according to the action sequence;

Specifically, the action basic element corresponding to each feature point in the current frame is obtained from the action sequence; and the position of each feature point in the current frame is calculated according to the parameter in the action basic element corresponding to each feature point in the current frame. Among them, the current frame refers to a frame that is about to arrive.

For example, first know the frame to which the current frame is going: Initialize the number of frames iTime=0 at the beginning of an action, iTime=iTime+l for each frame of animation, if iTime reaches the number of frames that the current action continues, Zero the iTime and start counting again.

After knowing the current frame is the first few frames, the position of each feature point in the current frame is calculated, and the feature point is j:

For a panning primitive, the amount of translation is:

(strTrans [ j] . iX * iTime I strTrans [j] . iTime,

strTrans [ j] . iY * iTime I strTrans [ j] . iTime);

See the definition of the data structure described above, where strTrans [j] . iX represents the total displacement of the feature point j in the X direction, strTrans [j] . iY represents the total displacement of the feature point j in the y direction, strTrans [j] . iTime represents the number of frames in which the feature point j is shifted, and iTime represents the number of frames in the current frame;

For the rotation primitive, first calculate the rotation vector V0, and then calculate the vector VI rotated on the two-dimensional plane of the image, the rotation angle is strRotate [j] . fTheta * iTime I strRotate [ j] . iTime, and then calculate The vector V2 after the rotation of the two-dimensional plane is perpendicular, and the rotation angle is strRotate [j] . fZ * iTime I strRotate [j] . iTime, the rotation base point plus V2 is the new position of the feature point j; The definition of the data structure described above;

For the convergence primitive, first calculate the total motion vector V0, and then get the motion vector of the current frame VI = V0 * strMix [j] . fRate * iTime I strMix [j] . iTime, where the meaning of each parameter is as described above The definition of the data structure.

In the above calculation, if iTime is greater than the number of frames in which the action elementary element continues, iTime is equal to the number of frames in which the action elementary element continues.

The position of the current frame feature point is calculated before each frame starts. When calculating, some feature points are associated, and the position of the associated feature point is also recalculated, as shown in Figure 4 When the feature point 8 is rotated, the feature point 12 also needs to be rotated at the same time. In the example shown in FIG. 4, the associated feature point group includes:

The set of feature points associated in translation: 6 ( 8, 10, 12), 7 ( 9, 11, 13 ) The set of feature points associated in the rotation: 8 ( 10, 12), 9 ( 11, 13 )

The set of feature points associated in the meeting: 10 ( 12), 11 ( 13), 12 ( 10), 13 ( 11 )

804: Deform the image of the body region in the image according to the initial position of the feature point and the position of the feature point in the current frame;

Optionally, the feature points of the shape in the image may be restored to the initial position according to the initial position of the acquired feature points, and each frame in the animation is deformed by using the initial position state, and the state of the initial position is The frame animation is obtained by image deformation before.

Further, the image of the shape of the image in the image is deformed according to the initial position of the feature point and the position of the feature point of the current frame, and specifically includes: connecting the feature point according to the initial position of the feature point and the relationship between the feature points Obtaining an initial feature line; connecting the feature points according to the relationship between the position of the feature points in the current frame and the feature points, to obtain the current frame feature line; according to the initial feature line and the current frame feature line, each shape region in the image The image is image-based based on feature line distortion. The reason why the feature line based image deformation technique is employed in the embodiment of the present invention is that the movement of the human body and the animal is driven by the bone.

In addition, since the image deformation has continuity, it is not necessary to calculate every pixel at the time of deformation, and the information of the uncalculated pixel points can be obtained by interpolation from the already calculated pixels.

805: An image of the deformed body region is overlaid on the background image in the image to generate an animation.

The above 801 to 805 are processing flow for generating one frame of animation, and then 802 to 805 are executed cyclically when the next frame animation is generated, and a continuous body animation can be generated by continuously looping.

When performing feature line-based image deformation, for a relatively single feature line, the area farther from the feature line and the area with a larger deformation range are prone to deformation and diffusion, for example, when the arm rotates more than 90 degrees during the test. There will be a phenomenon of deformation and diffusion. In order to prevent deformation and diffusion when the image is deformed, an auxiliary feature line may be added on the outer side of the body region, that is, the contour of the body body, and the shape is surrounded by the auxiliary feature line (may not be completely surrounded). The shape inside the auxiliary characteristic line is a phenomenon in which deformation and diffusion do not occur, and on the outside, a region which is relatively close to the auxiliary characteristic line is also a phenomenon in which deformation and diffusion do not occur.

For example, referring to Figure 9, the characteristic lines of the body region are 0-1, 1-2, 1-3, 2_4, 3_5, 0-6, and 0_7, and the characteristic lines of the left-hand region are 6-8, 8-10, and 10- 12, and copy the feature lines of the left-hand area to the two sides to obtain new feature lines 6, -8 ', 6 "-8", 8, -10, 8"-10", 10, -12, And 10"-12", the characteristic lines of the right-hand area are 7-9, 9-11, and 11-13, and the characteristic lines of the right-hand area are copied to each side to obtain new characteristic lines 7, -9, , 7 "-9" , 9, -11, 9" -11 ", 11, -13, and 11 " -13 ". For the copy operation of the left-hand area feature line, first calculate the length L1 of 6-8, and take 1^=1^1/5, where L is the distance between the copied feature line and the original feature line. Then, for each feature line to be copied on each segment of the left-hand region, the feature line is translated L in two directions perpendicular thereto, and each can be Get two new feature lines. Then, the intersection point of the adjacent feature line on the same side is obtained, and the intersection point is the intersection point of the new feature line corresponding to the intersection point of the original feature line. For example, 6-8, 8-10, 10-12 can be copied to one side to obtain 6'. -8—1, 8—2-10—1, 10—2-12, , and then find the intersection of 6′ -8—1 and 8—2-10—1, and get 8′, find 8—2 The intersection of -10-1 and 10-2-12, you can get 10,.

In the embodiment of the present invention, the shape of the body in the original input image input by the user or the system may be any posture, and is not specifically limited. The shape and shape region of the posture can be performed by using the method provided by the embodiment of the present invention. The positioning, as well as the predefined sequence of actions for it, generates a body animation. However, in order to make the body animation more smooth, more action, better effect, and simple pre-definition of the action sequence, the original input image that can be input is an image of a body standing upright and naturally drooping, if the input is not the same The image of the posture may be deformed into an image of the natural standing of the standing hands by the method provided by the embodiment of the present invention, and the deformed image is saved as a default image for subsequent deformation when the animation is generated, and according to This default image generates an action sequence.

The method provided by the embodiment of the present invention locates a feature point and a body region of a shape by inputting an image containing a body or an animal, and performs image repair on the image of the body region to repair the background image. As the background of the animation, the feature point position of each frame is calculated according to the sequence of motions read, and the image deformation of one frame and one frame is performed based on the initial position of the feature point, thereby generating a two-dimensional shape animation, which is adopted by the method The two-dimensional image deformation technique calculates a new position of a feature point in each frame according to the action sequence, forms a feature line, and then deforms to form an image in a new frame, thereby generating an animation effect without requiring a three-dimensional model, thereby reducing Workload; This method proposes four motion basic elements of two-dimensional animation, which are used to combine to form motion and motion sequences, and to drive motion through the motion sequence. Therefore, it is only necessary to modify the motion sequence to drive a single image to form an arbitrary motion. The animation of the action, without the need for a lot of The shape motion in the image is analyzed and clustered to obtain different types of motions. Therefore, the method is simple to implement and the amount of calculation is small. In addition, the operation of shifting the basic element can also produce some exaggerated motion effects in the two-dimensional animation, such as The feature points of the shoulder mentioned in the above embodiment are suddenly shifted outward to achieve the effect that the body suddenly becomes strong. Embodiment 3

Embodiments of the present invention provide an apparatus for generating a form animation, which can generate an animation of a form in an image for any given image, wherein the shape in the image can be in a human body, an animal, or a cartoon image. One kind. The device can be used to generate upper body animations of human bodies, animals, cartoons, etc., as well as full-body animations. Referring to Figure 10, the device includes:

The obtaining module 1001 is configured to acquire an initial position of a feature point of the shape in the image; The calculation module 1002 is configured to read an action sequence of the feature point, and calculate a position of the feature point in the current frame according to the action sequence;

The deformation module 1003 is configured to perform image deformation on the image of the shape region in the image according to the initial position of the feature point acquired by the acquisition module 1001 and the position of the feature point in the current frame calculated by the calculation module 1002;

The generating module 1004 is configured to cover an image of the deformed region of the deforming module 1003 on the background image in the image to generate an animation.

Further, referring to FIG. 11, the device further includes:

The first pre-defined module 1005 is configured to scan the original input image before acquiring the initial position of the feature point of the shape in the image, to obtain a feature point and a body region predefined for the shape in the image; For the specific implementation process of the module 1005, refer to step 201 in the second embodiment, and details are not described herein.

The processing module 1006 is configured to display the feature points and the body regions predefined by the first pre-defined module 1005 into the image, and save the shape of the predefined feature points and the body regions displayed by the user into the image. The position of the corresponding position on the corresponding location is performed. The specific implementation process of the processing module 1006 is shown in steps 202 and 203 in the second embodiment, and details are not described herein again.

Further, referring to FIG. 12, the apparatus further includes:

The repairing module 1007 is configured to perform image restoration on the image according to the position of the physical region accurately positioned by the user saved by the processing module 1006, and obtain an image and a background image of each physical region. The specific implementation process of the repairing module 1007 is detailed. Step 204 in Example 2, and details are not described herein again.

Further, referring to FIG. 13, the device further includes:

The second pre-defined module 1008 is configured to pre-define an action basic element before the calculation module 1002 reads the action sequence of the feature point, where the action basic element is used to indicate a change in the position of the feature point and a number of frames in which the change lasts. The element includes a static basic element, a translation basic element, a rotation basic element, and a convergence basic element. For the specific implementation process of the second pre-defined module 1008, refer to step 701 in the second embodiment, and details are not described herein.

The third pre-defined module 1009 is configured to pre-define the action according to the action basic element predefined by the second pre-defined module 1008 and the feature point of the feature, where each feature point corresponds to one action basic element; the third predefined For the specific implementation process of the module 1009, refer to step 702 in the second embodiment, and details are not described herein again.

The combination module 1010 is configured to combine the actions defined by the third pre-defined module 1009 into the action sequence of the feature points. For the specific implementation process of the combination module 1010, refer to step 703 in the second embodiment, and details are not described herein again.

Specifically, the third pre-defined module 1009 is specifically configured to confirm the position change of the feature point of the feature in the action one by one, and when the position of the feature point is confirmed to be unchanged, the position change of the feature point is represented by a static basic element. When confirming that the feature point is to be translated, the position change of the feature point is represented by a translation basic element, and the parameter of the translation basic element includes the The two-dimensional displacement of the feature point movement and the number of frames that continue to move; when it is confirmed that the feature point is to be rotated around other feature points, the position change of the feature point is represented by a rotation basic element, and the parameters of the rotation basic element include the surrounded feature Point, the angle of rotation on the two-dimensional plane of the image, the angle of rotation in the direction of the two-dimensional plane, and the number of frames in which the rotation continues; when it is confirmed that the feature point is to move to other feature points, the position of the feature point is changed Represented by the convergence element, the parameters of the convergence element include the target feature point, the movement ratio, and the number of frames that are moved.

Further, the calculating module 1002 includes:

An acquiring unit, configured to obtain, from the action sequence, an action basic element corresponding to each feature point in the current frame, where the current frame refers to a frame to be reached;

And a calculating unit, configured to calculate a position of each feature point in the current frame according to a parameter in an action basic element corresponding to each feature point in the current frame acquired by the acquiring unit.

For the specific implementation process of the calculation module 1002, refer to step 803 in the second embodiment, and details are not described herein again.

Optionally, the obtaining module 1001 is configured to obtain a position of a feature point of the shape in the original input image, and use the position of the obtained feature point as an initial position of the feature point; or

The obtaining module 1001 is specifically configured to acquire the position of the feature point of the shape in the image saved when the previous motion is completed, and take the position of the acquired feature point as the initial position of the feature point.

For the specific implementation process of the obtaining module 1001, refer to step 802 in the second embodiment, and details are not described herein again.

Further, the deformation module 1003 is specifically configured to connect the feature points according to the relationship between the initial position of the feature point and the feature point to obtain an initial feature line; according to the position of the feature point in the current frame and the association between the feature points The feature points are connected to obtain the current frame feature line; according to the initial feature line and the current frame feature line, the image of each shape region in the image is subjected to feature line-based image deformation. For the specific implementation process of the deformation module 1003, refer to step 804 in the second embodiment, and details are not described herein again.

In summary, the embodiment of the present invention locates a feature point and a body region of a shape by inputting an image containing a body or an animal, and performs image repair on the image of the body region to repair the background. Image As the background of the animation, the feature point position of each frame is calculated according to the sequence of motions read, and the image deformation of one frame and one frame is performed based on the initial position of the feature point, thereby generating a two-dimensional shape animation, which is adopted by the method The two-dimensional image deformation technique calculates a new position of a feature point in each frame according to the action sequence, forms a feature line, and then deforms to form an image in a new frame, thereby generating an animation effect without requiring a three-dimensional model, thereby reducing Workload; This method proposes four motion basic elements of two-dimensional animation, which are used to combine to form motion and motion sequences, and to drive motion through the motion sequence. Therefore, it is only necessary to modify the motion sequence to drive a single image to form an arbitrary motion. The animation of the motion does not need to analyze and cluster the body motions in a large number of images to obtain different types of motions as in the prior art, so the method is simple to implement and the amount of calculation is small; in addition, the operation of shifting the basic elements is Can produce the effect of some exaggerated actions in a two-dimensional body animation, such as the above The feature points of the shoulder mentioned in the example are suddenly translated outward to achieve the effect that the body suddenly becomes strong. It should be noted that, in the device for generating the body animation provided by the above embodiment, only the division of each functional module described above is illustrated in the process of generating the body animation. In practical applications, the functions may be assigned to different functional modules according to needs. Upon completion, the internal structure of the device is divided into different functional modules to perform all or part of the functions described above. In addition, the device for generating the body animation provided by the above embodiment is the same as the method for generating the body animation. The specific implementation process is described in detail in the method embodiment, and details are not described herein again.

The serial numbers of the embodiments of the present invention are merely for the description, and do not represent the advantages and disadvantages of the embodiments.

A person skilled in the art may understand that all or part of the steps of implementing the above embodiments may be completed by hardware, or may be instructed by a program to execute related hardware, and the program may be stored in a computer readable storage medium. The storage medium mentioned may be a read only memory, a magnetic disk or an optical disk or the like.

The above is only the preferred embodiment of the present invention, and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., which are within the spirit and scope of the present invention, should be included in the protection of the present invention. Within the scope.

Claims

Claim

A method for generating a form animation, the method comprising:

Obtaining an initial position of a feature point of the shape in the image;

The method according to claim 1, wherein before the obtaining the initial position of the feature point of the shape in the image, the method further comprises:

Scanning the original input image to obtain a feature point and a body region predefined for the shape in the image;

Displaying the predefined feature points and body regions into the image;

The position where the predefined feature points and the body regions displayed by the user are dragged to the corresponding positions on the shape in the image is saved.

3. The method according to claim 2, wherein the user saves the displayed predefined feature points and body regions to a corresponding position on the shape in the image for precise positioning. The method further includes: performing image restoration on the image according to the saved position of the body region accurately positioned by the user, to obtain an image and a background image of each body region.

4. The method according to claim 1, wherein before the sequence of actions of the feature points is read, the method further comprises:

Predefining an action element, the action element is used to indicate a change in a feature point position and a number of frames in which the change continues, and the action element includes a static element, a translation element, a rotation element, and a convergence element;

Defining an action according to the action element and the feature point of the shape, wherein each feature point corresponds to an action element in the action;

The predefined actions are combined into a sequence of actions of the feature points.

The method according to claim 4, characterized in that: according to the action element and the feature point of the body Predefined actions, including:

Confirming the positional changes of the feature points of the body in the action one by one,

When it is confirmed that the position of the feature point is unchanged, the position change of the feature point is represented by the stationary basic element; when it is confirmed that the feature point is to be translated, the position change of the feature point is described by Translating a basic element representation, wherein the parameter of the translation basic element includes a two-dimensional displacement of the feature point movement and a number of frames that are continuously moved;

When it is confirmed that the feature point is to be rotated around other feature points, the position change of the feature point is represented by the rotation basic element, and the parameter of the rotation basic element includes the surrounded feature point, in the image a rotation angle on a two-dimensional plane, a rotation angle in a direction perpendicular to the two-dimensional plane, and a number of frames in which the rotation continues;

When it is confirmed that the feature point is to be moved to other feature points, the position change of the feature point is represented by the convergence basic element, and the parameters of the convergence element include the target feature point, the movement ratio, and the number of frames that are continuously moved. .

The method according to claim 5, wherein calculating the position of the feature point in the current frame according to the action sequence comprises:

Obtaining an action basic element corresponding to each feature point in the current frame from the action sequence;

And calculating a position of each feature point in the current frame according to a parameter in the action basic element corresponding to each feature point in the current frame.

7. The method according to claim 1, wherein the initial position of the feature points of the shape in the image is obtained, including:

Obtaining a position of a feature point of the shape in the original input image, and taking the obtained position of the feature point as an initial position of the feature point;

Or,

The position of the feature point of the feature in the image saved at the completion of the previous action is obtained, and the acquired position of the feature point is taken as the initial position of the feature point.

The method according to claim 1, wherein the image of the body region in the image is image-deformed according to the initial position of the feature point and the position of the feature point in the current frame, including:

And connecting the feature points according to an initial position of the feature point and an association between the feature points to obtain an initial feature line;

Correlating the feature points according to the position of the feature points in the current frame and the feature points, and obtaining Pre-frame feature line;

Based on the initial feature line and the current frame feature line, the image of each body region in the image is subjected to feature line based image deformation.

9. A method according to any one of claims 1 to 8, wherein the form in the image is one of a human, an animal and a cartoon.

10. An apparatus for generating a form animation, wherein the apparatus comprises:

An obtaining module, configured to acquire an initial position of a feature point of a shape in the image;

The device according to claim 10, wherein the device further comprises:

a first pre-defined module, configured to: before the acquiring module acquires an initial position of a feature point of the shape in the image, scan the original input image to obtain a feature point and a body region predefined for the shape in the image;

a processing module, configured to display a feature point and a body region predefined by the first predefined module into the image, and save a predefined feature point and a body region displayed by the user into the image The position of the corresponding position on the shape to be precisely positioned.

The device according to claim 11, wherein the device further comprises:

And a repairing module, configured to perform image repair on the image according to the position of the shape region accurately stored by the user and saved by the processing module, to obtain an image and a background image of each shape region.

13. The device according to claim 10, wherein the device further comprises:

a second pre-defined module, configured to pre-define an action basic element before the calculation module reads the action sequence of the feature point, where the action basic element is used to indicate a change in a feature point position and a number of frames in which the change continues. The action basic element includes Static primitive, translation primitive, rotation primitive, and convergence primitive;

a third pre-defined module, configured to pre-define an action according to the action basic element predefined by the second predefined module and the feature point of the shape, where each feature point corresponds to one action basic element;

And a combination module, configured to combine the actions predefined by the third predefined module into a sequence of actions of the feature points.

The apparatus according to claim 13, wherein the third pre-defined module is specifically configured to confirm the position change of the feature point of the shape in the action one by one, when confirming the feature point When the position is unchanged, the position change of the feature point is represented by the static basic element; when it is confirmed that the feature point is to be translated, the position change of the feature point is represented by the translation basic element, The parameter of the panning element includes a two-dimensional displacement of the feature point movement and a number of frames of the moving duration; when it is confirmed that the feature point is to be rotated around other feature points, the position of the feature point is changed by the rotation. Meta-representation, the parameters of the rotating elementary element include a surrounding feature point, a rotation angle on a two-dimensional plane of the image, a rotation angle in a direction perpendicular to the two-dimensional plane, and a number of frames in which the rotation continues; When the feature point is to be moved to other feature points, the position change of the feature point is represented by the convergence basic element, and the parameters of the convergence basic element include Standard feature point, the ratio of movement and the moving continuous frames.

The device according to claim 14, wherein the calculating module comprises:

An acquiring unit, configured to acquire, from the sequence of actions, an action basic element corresponding to each feature point in the current frame; and a calculating unit, configured to use, according to the action basic element corresponding to each feature point in the current frame acquired by the acquiring unit Parameters that calculate the position of each feature point in the current frame.

The device according to claim 10, wherein the acquiring module is configured to acquire a position of a feature point of the shape in the original input image, and use the acquired position of the feature point as the feature point. Or initial position; or, obtain the position of the feature point of the feature in the image saved at the completion of the previous action, and take the acquired position of the feature point as the initial position of the feature point.

The device according to claim 10, wherein the deformation module is configured to connect the feature points according to an initial position of the feature point and an association between the feature points to obtain an initial Feature line; connecting the feature points according to the relationship between the position of the feature point in the current frame and the feature point to obtain a current frame feature line; according to the initial feature line and the current frame feature line, Feature line based image deformation is performed on images of the respective body regions in the image.

18. Apparatus according to any of claims 10 to 17, wherein the shape in the image is one of a human body, an animal and a cartoon.