CN104183000B - Full-automatic multi-source heterogeneous motion redirecting method of quasi-man character - Google Patents

Abstract

The invention provides a full-automatic multi-source heterogeneous motion redirecting method of a quasi-man character. The full-automatic multi-source heterogeneous motion redirecting method comprises the following steps: according to the joint point topological structure characteristics of a target bone, determining a middle bone, and carrying out normalization alignment processing to the middle bone, the target bone and a source bone; then, carrying out simplification processing, independently carrying out hip, chest and lobe child node matching to the simplified target bone and source bone with the simplified middle bone, and carrying out wrist and ankle matching; independently carrying out semantic information adding to the target bone and the source bone, and according to a selected matching strategy, carrying out node mapping processing from the source bone to the target bone; and transferring the node motion data of the source bone to the nodes of the target bone, reducing a motion track by the target bone according to the node motion data of the source bone, and carrying out multi-source motion fusion. The invention provides a method which gives consideration to the features of a quasi-man cartoon character and can realize a great quantity of automatic redirection, and multi-source motion is effectively integrated.

Description

Fully automatic multi-source heterogeneous motion redirection method for humanoid characters

technical field

The invention relates to the technical field of computer animation, in particular to a fully automatic multi-source heterogeneous motion redirection method for a humanoid character.

Background technique

Motion redirection takes human skeletal animation as the main research object, and aims to transfer the motion of the source character to the target character while maintaining the visual and constraint characteristics of its motion. Among them, if the topology of the source bone and the target bone are the same, but only the length of the bone is different, it is called isomorphic redirection; if the topology of the source bone and the target bone is different, it is called heterogeneous redirection. For isomorphic characters, motion can be easily transferred joint by joint due to the same topology, while heterogeneous reorientation is where the difficulty lies.

The main process of heterogeneous retargeting includes joint point mapping of source bones and target bones, transfer of motion data, and feature repair. The purpose of joint point mapping is to establish a one-to-one or one-to-many mapping relationship between the joint points of the source character and the target character. Each node will inherit the motion information of its mapping node, and its technical points lie in the data inheritance strategy and the processing of the difference in the pose of the skeleton calibration; in feature restoration, it mainly studies the reproduction of a certain overall or partial feature of the source character on the target character, The key points of its technology are feature extraction and feature restoration.

The existing skeletal joint point matching methods mainly rely on manual matching by users. For example, the well-known animation production software motion builder allows users to manually establish a mapping between the source bone and the target bone when processing motion redirection, and then based on the mapping relationship Transfer exercise data. In the field of research, automatic redirection has always been one of the difficulties in research. Semi-automatic redirection methods that require manual semantic calibration of source and target bones, transfer of heterogeneous bone data through intermediate skeletons, and automatic rooting node and divide the bone into multiple branches from the root to the end node to match one by one.

For humanoid animation characters, due to the creativity of artists in general design, the character bones are also diverse, and there will be big ears, long noses, wings, tails, three-fold legs, etc. that other creatures in nature do not have. feature. Therefore, for these roles, the semantic method needs to manually identify the key nodes, which obviously cannot achieve real-time redirection for various roles; the intermediate skeleton method depends to a certain extent on the design of the intermediate skeleton, if the middle is too simple, a large number of nodes cannot be matched, and if the skeleton Redundancy is likely to cause matching misplacement; the automatically extracted root node will also have a large error due to the local complexity of the skeleton. Therefore, none of the above methods can take into account the characteristics of humanoid animation characters, so that a large number of automatic redirections cannot be realized.

Therefore, the prior art still needs to be improved and developed.

Contents of the invention

The technical problem to be solved by the present invention is to provide a fully automatic multi-source heterogeneous motion redirection method for humanoid characters in view of the above-mentioned defects of the prior art, which can effectively solve the problem that the prior art does not take into account the characteristics of humanoid animation characters , so as to achieve a large number of automatic redirection defects.

The technical solution adopted by the present invention to solve technical problems is as follows:

A fully automatic multi-source heterogeneous motion redirection method for a humanoid character, wherein the method includes the steps of:

A. Determine the intermediate bone according to the topological structure characteristics of the joint points of the target bone, and perform normalized alignment processing on the intermediate bone, the target bone and the source bone;

B. Simplify the normalized and aligned intermediate bones, target bones, and source bones, and match the simplified target bones and source bones with the simplified intermediate bones for hip, chest, and leaf nodes respectively. And according to the simplified matching results, match the normalized and aligned target bones and source bones with the intermediate bones for wrist and ankle respectively;

C. Add semantic information to the target bone and the source bone, and perform node mapping from the source bone to the target bone according to the selected matching strategy;

D. Transfer the node motion data of the source bone to the node of the target bone, and the target bone restores the motion trajectory according to the node motion data of the source bone, and performs multi-source motion fusion.

The fully automatic multi-source heterogeneous motion redirection method for humanoid characters, wherein the step A specifically includes:

A1. Determine the intermediate bone according to the topological structure characteristics of the joint points in the target bone;

A2. Scale and place the intermediate bone in the first bounding box whose length and width are unit lengths, and scale the height and width in the same proportion to obtain a normalized intermediate bone; scale and place the target bone in length and width In the second bounding box with unit length, height and width scaled in the same proportion, the normalized target bone is obtained; the source bone is scaled and placed in the third bounding box whose length and width are unit length, height and width scaled in the same proportion In the bounding box, the normalized source bone is obtained;

A3. Place the first bounding box, the second bounding box, and the third bounding box in a space Cartesian coordinate system, and make the bottom surface of the first bounding box, the bottom surface of the second bounding box, and the The bottom surfaces of the third bounding box are all aligned with the xoz plane, so that the right side of the first bounding box, the right side of the second bounding box, and the right side of the third bounding box are all aligned with the yoz plane, so that The front of the third bounding box, the front of the second bounding box, and the front of the third bounding box are all aligned with the xoy plane.

The fully automatic multi-source heterogeneous motion redirection method for humanoid characters, wherein the step B specifically includes:

B1. Delete all nodes with a degree of 2 and their two adjacent edges in the normalized intermediate bone, the normalized target bone, and the normalized source bone, and add another edge to connect the two adjacent edges points until only leaf nodes with degree 1 and fork nodes with degree greater than 2 are left to obtain simplified intermediate bones, simplified target bones and simplified source bones;

B2. According to the matching algorithm with the minimum total cost of connectivity constraints, the simplified target bone and the simplified source bone are respectively matched with the simplified intermediate bone for hip, chest and leaf nodes;

B3. Match the normalized target bone and the normalized source bone with the normalized intermediate bone for wrist and ankle respectively.

In the fully automatic multi-source heterogeneous motion redirection method for a humanoid character, all nodes in the normalized source skeleton in the step B3 are mapped once.

In the fully automatic multi-source heterogeneous motion redirection method for humanoid characters, the intermediate skeleton is a standard motion capture character skeleton, the simplest intermediate skeleton, or an intermediate skeleton with tail wings extended.

In the fully automatic multi-source heterogeneous motion redirection method for humanoid characters, the intermediate skeleton is an expandable intermediate skeleton.

A fully automatic multi-source heterogeneous motion redirection method for a humanoid character provided by the present invention, the method includes: determining an intermediate bone according to the topological structure characteristics of the joint points of the target bone, and combining the intermediate bone, the target bone, and the source bone Perform normalized alignment processing; simplify the intermediate bones, target bones, and source bones that have undergone normalized alignment processing, and compare the simplified target bones and simplified source bones with the simplified intermediate bones respectively. Match the leaf node, and match the normalized and aligned target bone and source bone with the middle bone respectively for the wrist and ankle according to the simplified matching result; add semantic information to the target bone and the source bone respectively, And according to the selected matching strategy, perform the node mapping process from the source bone to the target bone; transfer the node motion data of the source bone to the node of the target bone, and the target bone is based on the node motion data of the source bone Restore the motion trajectory and perform multi-source motion fusion. The invention provides a method for realizing a large amount of automatic redirection while taking into account the characteristics of humanoid animation characters, and effectively integrates multi-source motions.

Description of drawings

Fig. 1 is a flow chart of a preferred embodiment of a fully automatic multi-source heterogeneous motion redirection method for a humanoid character according to the present invention.

Fig. 2 is a specific flow chart of normalization alignment processing in the fully automatic multi-source heterogeneous motion redirection method for humanoid characters described in the invention.

Figure 3a, Figure 3b, and Figure 3c are schematic diagrams of the standard motion capture character skeleton, the simplest intermediate skeleton, and the intermediate skeleton with tail wings extended in the present invention, respectively.

Fig. 4 is a schematic diagram of bone normalization in the fully automatic multi-source heterogeneous motion redirection method for humanoid characters described in the invention.

Fig. 5 is a specific flow chart of simplification and matching processing in the fully automatic multi-source heterogeneous motion redirection method for humanoid characters according to the present invention.

Fig. 6 is a schematic diagram of skeleton semantic markup in the present invention.

Figure 7a and Figure 7b are schematic diagrams of the source bone and target bone topology respectively, Figure 7c is a schematic diagram of mapping matching between the head of the source bone and the head of the target bone, Figure 7d is a schematic diagram of mapping matching between the right arm of the source bone and the right arm of the target bone, Figure 7e is a schematic diagram of mapping and matching between the right finger of the source bone and the right finger of the target bone, Figure 7f is a schematic diagram of mapping and matching between the spine of the source bone and the spine of the target bone, and Figure 7g is a schematic diagram of the mapping and matching of the right leg of the source bone and the right leg of the target bone, Fig. 7h is a schematic diagram of mapping matching between the right toe of the source bone and the right toe of the target bone, and FIG. 7i is a schematic diagram of mapping matching between the tail of the source bone and the tail of the target bone.

Fig. 8 is a structural block diagram of a preferred embodiment of a fully automatic multi-source heterogeneous motion redirection system for humanoid characters according to the present invention.

detailed description

In order to make the object, technical solution and advantages of the present invention more clear and definite, the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention.

Please refer to FIG. 1 . FIG. 1 is a flowchart of a preferred embodiment of a fully automatic multi-source heterogeneous motion redirection method for a humanoid character according to the present invention. As shown in Figure 1, the fully automatic multi-source heterogeneous motion redirection method of the humanoid character comprises the following steps:

Step S100: Determine an intermediate bone according to the topological structure characteristics of joint points of the target bone, and perform normalized alignment processing on the intermediate bone, the target bone, and the source bone.

In the embodiment of the present invention, firstly, a bone with motion action is selected as a source bone from a motion library storing a plurality of motion characters, and then a bone without motion motion is selected as a target bone. Then, the intermediate bone is determined according to the topological structure characteristics of the joint points of the target bone, and the source bone, intermediate bone and target bone are normalized and aligned, and the intermediate bone is used as an intermediary for matching the source bone and the target bone. By selecting an intermediate bone as the medium for matching the two, the matching is more accurate.

Step S200: Simplify the normalized and aligned intermediate bones, target bones, and source bones, and match the simplified target bones and the simplified source bones with the simplified intermediate bones for hip, chest, and leaf nodes respectively , and match the normalized and aligned target bone and source bone with the intermediate bone for wrist and ankle respectively.

After the source bone, the intermediate bone and the target bone are all normalized and aligned in step S100, they cannot be directly matched. Instead, the normalized alignment-processed intermediate bones, target bones, and source bones are simplified first, and then the source bones and target bones are matched with the intermediate bones at the hip, chest, and leaf nodes respectively, and according to the simplified matching results The target bone and source bone that have only undergone normalized alignment are matched with the intermediate bone for wrist and ankle respectively.

Step S300 , add semantic information to the target bone and the source bone, and perform node mapping from the source bone to the target bone according to the selected matching strategy.

Step S400, transfer the node motion data of the source bone to the node of the target bone, and the target bone restores the motion track according to the node motion data of the source bone, and performs multi-source motion fusion.

As a further embodiment, as shown in FIG. 2 , the normalized alignment processing in step S100 specifically includes:

Step S101. Determine an intermediate bone according to the topological structure characteristics of the joint points in the target bone.

The criteria for selecting the intermediate bone is to cover as many forked nodes as possible of the target bone and ensure that these nodes match. Although the intermediate skeleton is used in the embodiment of the present invention, experiments have proved that the intermediate skeleton only needs to have a basic semantic structure to obtain good results when it only participates in the matching of the first two layers of skeletons. Therefore, in most cases, you can choose the standard motion capture character skeleton as the intermediate skeleton (including six possible bifurcated nodes including hip, chest, two wrists, and two ankles). Angel's wings, etc.) for semantic description, the corresponding nodes can be added appropriately. During specific implementation, as shown in Figures 3a-3c, the intermediate skeleton can be selected from the standard motion capture character skeleton as shown in Figure 3a, the simplest intermediate skeleton as shown in Figure 3b, or the extended tail and wings as shown in Figure 3c Select one of the middle bones as the middle bone. The intermediate skeleton is not limited to the above three intermediate skeletons, and is an expandable intermediate skeleton. The intermediate bone does not require a unique structure, but can be expanded according to the shape of the source and target bones to increase adaptability.

Step S102. Scale and place the intermediate bone in the first bounding box whose length and width are both unit lengths, and whose height and width are scaled in the same proportion to obtain a normalized intermediate bone; scale and place the target bone in a long In the second bounding box whose width is unit length, height and width are scaled in the same proportion, the normalized target bone is obtained; the source bone is scaled and placed in the second bounding box whose length and width are unit length, height and width are scaled in the same proportion In the three bounding boxes, the normalized source bones are obtained.

The characters to be matched are all kinds of strange and different in size, which will have a great impact on node matching. All characters need to be normalized and aligned first. As shown in FIG. 4 , it is a schematic diagram of bone normalization in the fully automatic multi-source heterogeneous motion redirection method for a humanoid character according to the present invention. Firstly, it is stipulated that the top of the head of all characters points to the positive direction of the y-axis and faces the positive direction of the z-axis; secondly, all characters are scaled so that the width of the bounding box in the direction of the x-axis and y-axis is 1, and the direction of the z-axis is in the same proportion as the y-axis zoom. In step S102, the intermediate bone, the target bone and the source bone are respectively normalized so that they are at the same size level, which is more conducive to node matching.

Step S103, placing the first bounding box, the second bounding box, and the third bounding box in the spatial rectangular coordinate system, and making the bottom surface of the first bounding box, the bottom surface of the second bounding box, and the bottom surface of the third bounding box The bottom surface of the third bounding box is all aligned with the xoz plane, so that the right side of the first bounding box, the right side of the second bounding box, and the right side of the third bounding box are all aligned with the yoz plane, Align the front of the third bounding box, the front of the second bounding box, and the third bounding box with the xoy plane.

In a further embodiment, as shown in FIG. 5, the simplification and matching processing in step S200 specifically includes:

Step S201, delete all nodes with degree 2 and their two adjacent edges in the normalized intermediate bone, the normalized target bone and the normalized source bone, and add an edge to connect the two Neighboring nodes until there are only leaf nodes with a degree of 1 and fork nodes with a degree greater than 2, and the simplified intermediate bone, simplified target bone and simplified source bone are obtained.

For each node of the simplified intermediate bone, the mapping node of the displacement is specified in the simplified character to be matched (source bone or target bone), and the mapping nodes of different nodes cannot be the same.

Step S202 , according to the matching algorithm with the minimum total cost of connectivity constraints, the simplified target bone and the simplified source bone are respectively matched with the simplified intermediate bone at the hip, chest and leaf nodes.

For all adjacent node pairs in the simplified intermediate bone, after mapping to the bone to be matched, if other mapped nodes are included in the path connecting the mapped node pair, it is considered that it does not satisfy the connectivity constraint, and the node is removed. plan;

Second, for a mapping scheme that satisfies connectivity constraints, calculate the Euclidean distances between all nodes and their mapping nodes and sum them up as the total cost (distance) of the scheme.

Select the match that satisfies the connectivity constraints and the minimum total cost, and determine the hip and thoracic nodes of the bones to be matched. At the same time, the matching relationship of a group of leaf nodes can be obtained to guide the next step.

Step S203 , matching the normalized target bone and the normalized source bone with the normalized intermediate bone for wrist and ankle respectively.

In the embodiment of the present invention, in addition to matching the hip, chest and leaf nodes, it is also necessary to match the wrist and ankle nodes. The matching in this step is to match between the unsimplified normalized intermediate bone, the normalized target bone and the normalized source bone.

Take the left hand as an example: according to the mapping relationship obtained in step S202, find the mapping node of the left-hand leaf node of the middle bone. Find all subtrees rooted at the child node of the skeleton chest node to be matched. If the subtree contains the above mapping node, it is considered as a left-hand subtree, and the path from the root node to each leaf node in the left-hand subtree is regarded as a path , for each node of each path, calculate the distance function as follows: length ratio, that is, the ratio of the length of the path from the root node to the node to the total length of the path; area ratio, that is, the sum of the path length from the node to the adjacent node and the total length of the path The ratio of ; degree, that is, the degree of the node. The weighted average of the three values is carried out, and the weight is (2,1,1), and then the point with the smallest distance from the left wrist of the middle node is selected among all the results as the left wrist of the character to be matched. The right wrist is calculated in the same way, and the left leg and right leg are also calculated in the same way, the difference is that the root node of the left and right leg subtrees is selected from the child nodes of the hip node.

As a further embodiment, as shown in FIG. 6 , which is a schematic diagram of bone semantic markup in the present invention, in the step S300, when adding semantic information to the target bone and the source bone, it is not simplified Semantic information is attached to the target bone and the simplified source bone, but in step S200 after matching the hip, chest and leaf nodes and the matching of the wrist and ankle, after determining the above key nodes of the source bone and the target bone, According to a preset bone semantic tagging rule, the following sequence of tagging is performed:

Left/right fingers: Each subtree with each child node of the left/right wrist as the root node is a finger, and the z-axis coordinates of its leaf nodes are sorted into fingers one, two, three...; as shown in Figure 6 , the right finger one 2110 , right finger two 2120 and right finger three 2130 left fingers on the subtree with the right wrist 2100 as the root node, and the left finger one 2210 and left finger on the subtree with the left wrist 2200 as the root node Two 2220 and left finger three 2230;

Left/right arm: all nodes on the pathway from the chest node to the left/right wrist node (excluding the chest and wrist); All nodes on the path to left wrist 2200 are left arm 2400;

Head: Each subtree rooted at the unmarked sub-node of the chest node is a head, and it is sorted as head one, two, three... according to the horizontal distance from its leaf node to the chest node; as shown in Figure 6, Each subtree rooted at the unmarked child node of chest 1000 is marked as head 1 1100, head 2 1200 and head 3 1300, among which the horizontal distance from the leaf node of head 1100 to chest 1000 is the shortest, and head 2 1200 Second, the first three are 1300 the largest;

Left/right toe: Each subtree with each child node of the left/right ankle as the root node is a toe, and it is sorted as toe one, two, three... according to the increasing distance from the leaf node to the hip on the x-axis; as shown in the figure 6, the right toe one 4110 and the right toe two 4120 on the subtree with the right ankle 4100 as the root node, and the left toe one 4210 and the left toe two 4220 on the subtree with the left ankle 4200 as the root node;

Left/right leg: all nodes on the pathway from hip node to left/right ankle node (not including hip and ankle); as shown in Figure 6, all nodes on the pathway from hip 3000 to right ankle All nodes on the path from 3000 to left ankle 4200 are left leg 3200;

Spine: the path with the child node of the hip node as the root and the chest node as the leaf; as shown in Figure 6, the path with the child node of the hip 3000 as the root and the leaf of the chest 1000 is the spine 1400;

Tail: Each subtree rooted at the unmarked child node of the hip node is a tail, which is sorted according to the increasing horizontal distance from the longest leaf node to the hip node as tail 1, 2, 3...; as shown in Figure 6 , each subtree rooted at the unmarked child node of Hip 3000 is marked as Tail 1 3010, Tail 2 3020 and Tail 3 3030, among which the horizontal distance from the leaf node of Tail 1 3010 to Hip 3000 is the shortest, and Tail 2 3020 is the second, and the last three is 3030 the largest;

Other attachments: All unmarked nodes inherit the semantics of their parent nodes. If you need to support special semantics such as wings, you can extend the middle bone for semantic marking.

In the step S300, the matching strategy used for the node mapping process from the source bone to the target bone includes a median 1-1 matching strategy, a median 1-n matching strategy, a nearest-end matching strategy and a serial number matching strategy, wherein:

The median 1-1 matching strategy is to match the two branches of the selected source character and the target character at the median (it can be the spatial position, the plane projection position or the median value of the coordinate axis value, the latter is usually used), applicable In the case of head and tail;

The median 1-n matching strategy is to match the branch at the median of the selected source role with the n branches of the target role, which is applicable to the case of head and tail;

The nearest-end matching strategy is to match all branches of the target character with the closest source character branch of its leaf node, which is applicable to many situations such as head, tail, arm, hand, leg, foot, etc.;

The serial number matching strategy is when m≥n, match the n branches of the target bone with the first n branches of the source bone; when m<n, match the first m branches of the target bone with the m branches of the source bone Match in order, and match the m+1th to nth branches of the target bone with the mth branch of the source bone. This strategy is suitable for many situations such as head, tail, arm, hand, leg, foot and so on.

Generally, the median 1-1 matching is selected for the head, the median 1-n matching is selected for the tail, and the sequence number is selected for other parts to match, and the nearest end matching is used as a supplementary method when the semantics are unclear. In addition, other strategies can be appended. For selected matching branch pairs, the same mapping strategy as for single chains is used.

In step S300, the schematic diagrams of matching the target bone and the source bone according to the above matching principle are shown in Fig. Schematic diagram of the mapping and matching of the head of the target bone. Figure 7d is a schematic diagram of the mapping and matching of the right arm of the source bone and the right arm of the target bone. Figure 7e is a schematic diagram of the mapping and matching of the right finger of the source bone and the right finger of the target bone. Schematic diagram of the mapping and matching of the spine of the target bone. Figure 7g is a schematic diagram of the mapping and matching of the right leg of the source bone and the right leg of the target bone. Figure 7h is a schematic diagram of the mapping and matching of the right toe of the source bone and the right toe of the target bone. Schematic diagram of the mapping and matching of the skeletal tail.

As a further embodiment, in the step S400, the movement difference caused by the arms of the three poses of N T A is processed. For the nodes of the arms and fingers, each bone needs to be rotated to the same pose as the corresponding bone of the source character in the calibrated pose , after the action is transferred, it is rotated back to the original position to process its child nodes, and the other joint corners are directly transferred to the target character frame by frame. Among them, N-pose refers to the posture of standing upright with hands hanging down naturally, T-pose refers to the posture of standing upright with hands sideways and horizontally opened, and A-pose refers to standing upright with hands open sideways and the trunk at about 45 degrees attitude.

In step S400, when the target bone restores the motion track according to the node motion data of the source bone, first record the root node of the source action and the left and right foot tracks; then scale the track according to the height of the hip nodes of the source and target bones; Finally, the scaled trajectory is used as the mandatory trajectory of the target bone, and IK (Inverse Kinematic, that is, inverse kinematics) is calculated to obtain a new leg movement.

At the same time, when performing multi-source motion fusion in step S400, one of the cross strategy, combination strategy or mutation strategy is adopted, wherein:

The crossover strategy refers to the division of the bones of the target character into regions, and the actions of each region come from different source characters, thus combining them into new motions. For example, source character A has no wings and a tail, source character B has wings and source character C has a tail. At this time, for a target character T that has both wings and a tail, you can get the body movement from A, the wing movement from B, and C For example, for a source movement A of squatting and shooting and a source movement B of running, the upper and lower body movements can be obtained respectively, and the target movement T of running and shooting can be synthesized.

The mutation strategy refers to adding subtle perturbations to the source motion so that the target character has new motion characteristics. For example, adding a backward posture to the original ordinary walking action generates a backward walking movement; another example is to generate a joyful walking action by increasing the swinging range of the hands in the original running action.

Combination strategy refers to splicing multiple source motions to form a new set of target character motions. For example, for a running and a shooting action, by splicing and calculating the intermediate transition, an action sequence of running→shooting is generated.

Step S100 - step S400 can be completed automatically, wherein the intermediate skeleton and matching strategy can be set manually or automatically. Assume that the motion library has N motion characters, including m motion bones, s motions, and n non-motion bones. After completing the fully automatic multi-source heterogeneous motion redirection of the humanoid characters, each character in the motion library will have s actions and become a character motion library with N*s motion clips, opening up the multi-source relocation of heterogeneous characters. A key aspect of orientation. After that, multi-source motion fusion is performed on each character itself, and the fused motion can be applied to the rest of the characters after redirection, thereby realizing the expansion of the geometric level of the motion library.

Based on the above embodiments, the present invention also provides a fully automatic multi-source heterogeneous motion redirection system for humanoid characters, as shown in FIG. 8 , the fully automatic multi-source heterogeneous motion redirection system for humanoid characters includes:

The normalized alignment module 100 is configured to determine an intermediate bone according to the topological structure characteristics of the joint points of the target bone, and perform normalized alignment processing on the intermediate bone, the target bone and the source bone; the details are as described above.

The intermediate matching module 200 is used to simplify the normalized and aligned intermediate bones, target bones, and source bones, and compare the simplified target bones and the simplified source bones with the simplified intermediate bones for hip and chest Match the leaf nodes, and match the normalized and aligned target bones and source bones with the intermediate bones for wrist and ankle respectively; the details are as described above.

The mapping and matching module 300 is configured to add semantic information to the target bone and the source bone, and perform node mapping from the source bone to the target bone according to the selected matching strategy; the details are as described above.

The motion transmission module 400 is configured to transmit the node motion data of the source bone to the node of the target bone, and the target bone restores the motion track according to the node motion data of the source bone, and performs multi-source motion fusion; Specifically as above.

As a further embodiment, in the fully automatic multi-source heterogeneous motion redirection system for humanoid characters, the normalized alignment module 100 specifically includes:

The intermediate bone determination unit is configured to determine the intermediate bone according to the topological structure characteristics of the joint points in the target bone; the details are as described above.

A normalization unit is used to scale the intermediate bone and place it in the first bounding box whose length and width are unit lengths, and whose height and width are scaled in the same proportion to obtain a normalized intermediate bone; scale the target bone And put it in the second bounding box whose length and width are both unit length, and the height and width are scaled in the same proportion to get the normalized target bone; scale and place the source bone in which the length and width are unit length, and the height and width are scaled at the same time In the scaled third bounding box, the normalized source bone is obtained; the details are as described above.

an alignment unit, configured to place the first bounding box, the second bounding box, and the third bounding box in a space Cartesian coordinate system, and make the bottom surface of the first bounding box and the bottom surface of the second bounding box and the bottom surface of the third bounding box are all aligned with the xoz plane, so that the right side of the first bounding box, the right side of the second bounding box and the right side of the third bounding box are all aligned with the yoz plane Aligning, so that the front of the third bounding box, the front of the second bounding box, and the front of the third bounding box are all aligned with the xoy plane; specifically as described above.

As a further embodiment, in the fully automatic multi-source heterogeneous motion redirection system for humanoid characters, the intermediate matching module 200 specifically includes:

A simplified unit is used to delete all nodes with a degree of 2 and their two adjacent edges in the normalized intermediate bone, the normalized target bone, and the normalized source bone, and add another edge to connect the two adjacent points until only leaf nodes with degree 1 and fork nodes with degree greater than 2 are left to obtain simplified intermediate bones, simplified target bones and simplified source bones; the details are as described above.

The first matching unit is used to match the simplified target bone and the simplified source bone with the simplified intermediate bone respectively for hip, chest and leaf nodes according to the matching algorithm with the minimum total cost of connectivity constraints; the details are as described above.

The second matching unit is used to match the normalized target bone and the normalized source bone with the normalized intermediate bone for wrist and ankle respectively; the details are as described above.

As a further embodiment, in the fully automatic multi-source heterogeneous motion redirection system for humanoid characters, all nodes in the normalized source skeleton in the second matching unit are mapped once; specifically as described above.

As a further embodiment, in the fully automatic multi-source heterogeneous motion redirection system for humanoid characters, the intermediate skeleton is a standard motion-capture character skeleton, the simplest intermediate skeleton, or an intermediate skeleton with tail wings extended; specifically as described above stated.

To sum up, the present invention provides a fully automatic multi-source heterogeneous motion redirection method for a humanoid character, the method includes: determining an intermediate bone according to the topological structure characteristics of the joint points of the target bone, and combining the intermediate bone, the The target bone and the source bone are normalized and aligned; the normalized and aligned intermediate bone, the target bone and the source bone are simplified, and the simplified target bone and the simplified source bone are respectively compared with the simplified intermediate bone. Match the hip, chest and leaf nodes of the bones, and match the normalized and aligned target bones and source bones with the intermediate bones for wrist and ankle respectively; add semantic information to the target bones and the source bones respectively , and according to the selected matching strategy, perform the node mapping process from the source bone to the target bone; transfer the node movement data of the source bone to the node of the target bone, and the target bone moves according to the node movement of the source bone The data restores the motion trajectory and performs multi-source motion fusion. The invention provides a method for realizing a large amount of automatic redirection while taking into account the characteristics of humanoid animation characters, and effectively integrates multi-source motions.

It should be understood that the application of the present invention is not limited to the above examples, and those skilled in the art can make improvements or transformations according to the above descriptions, and all these improvements and transformations should belong to the protection scope of the appended claims of the present invention.

Claims (5)

1. A fully automatic multi-source heterogeneous motion redirection method for a humanoid character, characterized in that the method comprises steps: A. Determine the intermediate bone according to the topological structure characteristics of the joint points of the target bone, and perform normalized alignment processing on the intermediate bone, the target bone and the source bone; B. Simplify the normalized and aligned intermediate bones, target bones, and source bones, and match the simplified target bones and source bones with the simplified intermediate bones for hip, chest, and leaf nodes respectively. And according to the simplified matching results, match the normalized and aligned target bones and source bones with the intermediate bones for wrist and ankle respectively; C. Add semantic information to the target bone and the source bone, and perform node mapping from the source bone to the target bone according to the selected matching strategy; D. Transfer the node motion data of the source bone to the node of the target bone, and the target bone restores the motion track according to the node motion data of the source bone, and performs multi-source motion fusion; Described step B specifically comprises: B1. Delete all nodes with a degree of 2 and their two adjacent edges in the normalized intermediate bone, the normalized target bone, and the normalized source bone, and add another edge to connect the two adjacent edges points until only leaf nodes with degree 1 and fork nodes with degree greater than 2 are left to obtain simplified intermediate bones, simplified target bones and simplified source bones; B2. According to the matching algorithm with the minimum total cost of connectivity constraints, the simplified target bone and the simplified source bone are respectively matched with the simplified intermediate bone for hip, chest and leaf nodes; B3. Match the normalized target bone and the normalized source bone with the normalized intermediate bone for wrist and ankle respectively. 2. The fully automatic multi-source heterogeneous motion redirection method for humanoid characters according to claim 1, wherein the step A specifically includes: A1. Determine the intermediate bone according to the topological structure characteristics of the joint points in the target bone; A2. Scale and place the intermediate bone in the first bounding box whose length and width are unit lengths, and scale the height and width in the same proportion to obtain a normalized intermediate bone; scale and place the target bone in length and width In the second bounding box with unit length, height and width scaled in the same proportion, the normalized target bone is obtained; the source bone is scaled and placed in the third bounding box whose length and width are unit length, height and width scaled in the same proportion In the bounding box, the normalized source bone is obtained; A3. Place the first bounding box, the second bounding box, and the third bounding box in a space Cartesian coordinate system, and make the bottom surface of the first bounding box, the bottom surface of the second bounding box, and the The bottom surfaces of the third bounding box are all aligned with the xoz plane, so that the right side of the first bounding box, the right side of the second bounding box, and the right side of the third bounding box are all aligned with the yoz plane, so that The front of the third bounding box, the front of the second bounding box, and the front of the third bounding box are all aligned with the xoy plane. 3. The fully automatic multi-source heterogeneous motion redirection method for humanoid characters according to claim 1, characterized in that all nodes in the normalized intermediate skeleton in the step B3 are mapped once. 4. The fully automatic multi-source heterogeneous motion redirection method for humanoid characters according to claim 1, wherein the intermediate skeleton is a standard motion capture character skeleton, the simplest intermediate skeleton, or an intermediate skeleton with tail wings extended. 5. The fully automatic multi-source heterogeneous motion redirection method for a humanoid character according to claim 1, wherein the intermediate skeleton is an expandable intermediate skeleton.