A Motion Retrieval Method Based on GMM Semantic Features
A semantic feature and motion technology, applied in the field of 3D human motion data retrieval, can solve problems such as inability to quickly and accurately retrieve, and achieve accurate retrieval results
Active Publication Date: 2018-11-30
ZHEJIANG UNIV
View PDF6 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] The purpose of the present invention is to solve the problem that an animator cannot quickly and accurately retrieve the required motion when making animation, and a motion retrieval method based on GMM semantic features is provided and includes the following steps:
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0030] Such as figure 1 As shown, a motion retrieval method based on GMM semantic features includes the following steps:
[0031] 1) Carry out category labeling and training sample set division for any given 3D human motion data set, and perform rotation and translation alignment processing on each posture in all motion sequences in the data set, so that the center point is fixed as the coordinate origin, and the body plane The forward orientation is unified;
[0032] 2) extract key frame to the movement sequence in data set, its method is: to given movement sequence s={f 1 , f 2 ,..., f n}, where f i is a certain frame in the motion sequence, n is the total number of frames in the motion sequence, first use the k-means clustering algorithm to cluster all the frames, and divide the entire motion sequence according to the category number, the same continuous cluster number is A segment; then select a frame closest to the average position of the segment from each sub-segmen...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The present invention discloses a motion retrieval method based on GMM (Gaussian Mixture Model) semantic features. The retrieval method comprises: firstly, performing key-frame extraction on training data, and establishing a "key posture'' model based on a key frame by using a GMM, wherein the "key posture'' model is used for calculating semantic features for postures of all motion data; secondly, by statistically combing semantic features of posture granularity, generating semantic features of sequence granularity, wherein the semantic features of the sequence granularity are used for comparison between motion sequences; and finally, giving a to-be-retrieved motion sequence and features thereof, performing similar motion retrieval on a database by using a sparse coding method instead of a conventional K-nearest neighbor method, and listing out retrieval results according to the similarity. The method is accurate in retrieval effect and high in calculation efficiency, and meets the demand of quickly retrieving similar motions when an animator makes an animation.
Description
technical field [0001] The invention relates to a three-dimensional human motion data retrieval technology, a Gaussian mixture model and a sparse coding algorithm, in particular to a motion retrieval method using GMM semantic features. Background technique [0002] With the development of motion capture technology and the commercialization of motion capture devices such as multi-cameras and Microsoft Kinect depth cameras, human motion data has been widely used in various fields. In the related research in recent years, scholars tend to focus on the application of sports data, but the ever-increasing sports data itself is often ignored. Therefore, the current status quo is that animators collect or synthesize a large amount of human motion data when making animations, but rarely reuse existing motions reasonably in new movie or game productions. The main reason for this waste of resources is the lack of such a search engine that can effectively retrieve motion data from mass...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06F17/30
CPCG06F16/784
Inventor 肖俊齐天张翰之庄越挺
Owner ZHEJIANG UNIV