Human Motion Capture Based on Incremental Dimension Reduction and Projection Position Optimization

Wanyi Li,Guoming Chen,Yilin Wu,Qian Zhang,Yuqi Zeng,Giuseppe Piro

doi:10.1155/2021/5589100

Wanyi Li, Guoming Chen + Show 4 more

Open Access

https://doi.org/10.1155/2021/5589100

Copy DOI

Abstract

Three-dimensional (3D) human motion capture is a hot researching topic at present. The network becomes advanced nowadays, the appearance of 3D human motion is indispensable in the multimedia works, such as image, video, and game. 3D human motion plays an important role in the publication and expression of all kinds of medium. How to capture the 3D human motion is the key technology of multimedia product. Therefore, a new algorithm called incremental dimension reduction and projection position optimization (IDRPPO) is proposed in this paper. This algorithm can help to learn sparse 3D human motion samples and generate the new ones. Thus, it can provide the technique for making 3D character animation. By taking advantage of the Gaussian incremental dimension reduction model (GIDRM) and projection position optimization, the proposed algorithm can learn the existing samples and establish the relevant mapping between the low dimensional (LD) data and the high dimensional (HD) data. Finally, the missing frames of input 3D human motion and the other type of 3D human motion can be generated by the IDRPPO.

Highlights

Three-dimensional (3D) human motion capture is applied for many fields, such as medical diagnosis, animation making, and 3D video game development [1,2,3]
incremental dimension reduction and projection position optimization (IDRPPO) with the Gaussian incremental dimension reduction model (GIDRM) can help to learn the incomplete gait, and generate the other gait, which makes up the defects of some self-supervised or unsupervised algorithms
The experimental results can reveal IDRPPO is efficacious in making 3D human character animation, which can do great help to generating the motion cycle fast

Summary

Introduction

Three-dimensional (3D) human motion capture is applied for many fields, such as medical diagnosis, animation making, and 3D video game development [1,2,3]. E (updated) mapping f1 from incomplete motion cycle YI, the generated missing poses can XI toYIis built. The proposed method can obtain the new essential samples according to the data requirement of the self-supervised learning model and let the model update the generating mapping for the improvement of tracking or estimating by the help of these samples. The new algorithm (method) called incremental dimension reduction and projection position optimization (IDRPPO) is proposed to address the problems mentioned above It can generate one type human motion from the other type. GIDRM can provide the LD space for searching and generating the optimal LD data sample, so that the corresponding 3D human motion can be reconstructed by its mappings. The details of IDRPPO will be discussed

Generation of Human Motion through IDRPPO

Experiment and Evaluation

Conclusion