Mining Spatial-Temporal Patterns and Structural Sparsity for Human Motion Data Denoising

Abstract

Motion capture is an important technique with a wide range of applications in areas such as computer vision, computer animation, film production, and medical rehabilitation. Even with the professional motion capture systems, the acquired raw data mostly contain inevitable noises and outliers. To denoise the data, numerous methods have been developed, while this problem still remains a challenge due to the high complexity of human motion and the diversity of real-life situations. In this paper, we propose a data-driven-based robust human motion denoising approach by mining the spatial-temporal patterns and the structural sparsity embedded in motion data. We first replace the regularly used entire pose model with a much fine-grained partlet model as feature representation to exploit the abundant local body part posture and movement similarities. Then, a robust dictionary learning algorithm is proposed to learn multiple compact and representative motion dictionaries from the training data in parallel. Finally, we reformulate the human motion denoising problem as a robust structured sparse coding problem in which both the noise distribution information and the temporal smoothness property of human motion have been jointly taken into account. Compared with several state-of-the-art motion denoising methods on both the synthetic and real noisy motion data, our method consistently yields better performance than its counterparts. The outputs of our approach are much more stable than that of the others. In addition, it is much easier to setup the training dataset of our method than that of the other data-driven-based methods.

Human motion data refinement unitizing structural sparsity and spatial-temporal information

“…Following the work [5], [6], [9], [32], [33], the Root Mean Squared Error (RMSE) measurement is adopted to qualify the refined results:…”

“…In order to better exploit the spatial-temporal relationship, many researches have applied the partial human model while processing motion data [5], [6], [25], [27], [28] and achieved expressive performance. Instead of using whole body model [8], [9], We choose partial human model in this work and divide the whole body into 5 parts [5], [6], which is…”

Self‐similarity Analysis for Motion Capture Cleaning

“…We compared our approach with other recent methods; we chose methods that correct motion at the marker level, including matrix factorization [BL16], and methods that work at the joint level, including deep learning [HSKJ15], and sparse coding [FJX*15]. The experiments were implemented using data containing motion sequences of various locomotion (walk, run, jump), and dancing.…”

“…and Xiao et al . [FJX*15, XFJ*15] divide human pose into parts to learn multiple dictionaries that contain the spatial‐temporal patterns of human motion, that are later adopted to remove the noise and outliers from noisy data using sparse coding. Holden et al .…”

Action recognition based on global optimal similarity measuring

“…For the problem of unstable joints, Feng et al [5,6] explore how to refine and denoise motion capture data. Xiao et al [21] also provide an effective approach to denoise human motion data.…”