Studies on human motion have attracted a lot of attentions. Human capture data, which much more precisely records than videos do, has been widely used in many areas. Motion segmentation is an indispensable step for related applications, but current methods data do not effectively model some important characteristics such as Riemannian manifold structure and containing non-Gaussian noise. In this paper, we convert the into temporal subspace clustering problem. Under framework sparse...

A rapid detection method for heavy metals in oily soil is needed to provide accurate data support <italic>in situ</italic> pollution assessment and restoration.

Human motion capture data has been widely used in many areas, but it involves a complex process and the captured inevitably contains missing due to occlusions caused by actor's body or clothing. Motion recovery, which aims recover underlying complete sequence from its degraded observation, still remains as challenging task nonlinear structure kinematics property embedded data. Low-rank matrix completion-based methods have shown promising performance short-time-missing recovery problems....

In this paper, we propose an efficient and effective framework to fuse hyperspectral Light Detection And Ranging (LiDAR) data using two coupled convolutional neural networks (CNNs). One CNN is designed learn spectral-spatial features from data, the other one used capture elevation information LiDAR data. Both of them consist three layers, last layers are together via a parameter sharing strategy. fusion phase, feature-level decision-level methods simultaneously integrate these heterogeneous...

Human motion capture data, which are used to animate animation characters, have been widely in many areas. To satisfy the high-precision requirement, human data captured with a high frequency (120 frames/s) by system. However, and nonlinear structure make storage, retrieval, browsing of challenging problems, can be solved keyframe extraction. Current extraction methods do not properly model two important characteristics i.e., sparseness Riemannian manifold structure. Therefore, we propose...

In this paper, we propose a new GAN-based framework to implement video-based human motion transfer, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">i.e.</i> , transferring the motions from source person target one with help of pose information. Human transfer involves large scaled spatial deformations body image and emphasizes consistency details. However, GAN is not suitable for region-unaligned task due global adversarial loss does focus on...

Recently, sparse representation-based classification (SRC) has been widely studied and produced state-of-the-art results in various tasks. Learning useful computationally convenient representations from complex redundant highly variable visual data is crucial for the success of SRC. However, how to find best feature representation work with SRC remains an open question. In this paper, we present a novel discriminative projection learning approach objective seeking matrix such that learned...

Sparse representation-based classification (SRC) has been developed and shown great potential for real-world application. Based on SRC, Yang et al. devised an SRC steered discriminative projection (SRC-DP) method. However, as a linear algorithm, SRC-DP cannot handle the data with highly nonlinear distribution. Kernel sparse classifier (KSRC) is non-linear extension of can remedy drawback SRC. KSRC requires use predetermined kernel function selection its parameters difficult. Recently,...

By exploiting the kernel trick, sparse subspace model is extended to nonlinear version with one or a combination of predefined kernels, but high-dimensional space induced by kernels not guaranteed be able capture features data in theory. In this article, we propose nonconvex low-rank learning framework an unsupervised way learn replace model. The learned relaxation rank can better property induce Hilbert that more closely approaches true feature space. Furthermore, give global closed-form...

Motion synthesis technology can produce natural and coordinated motion data without a capture process, which is complex costly. Current methods usually provide few interfaces to avoid the arbitrariness of but this actually reduces understandability process. In paper, we propose learning-based Sphere nonlinear interpolation (Snerp) model that generate in-between motions in terms given start-end frame pair. Variety input pairs will enrich diversity generated motions. The angle speed human not...

Deep subspace learning is an important branch of self-supervised and has been a hot research topic in recent years, but current methods do not fully consider the individualities temporal data related tasks. In this paper, by transforming motion capture segmentation task as supervision, we propose local self-expression network. Specifically, considering temporality data, use convolution module to extract features. To implement validity tasks, design layer which only maintains representation...

A realistic 2-D motion can be treated as a deforming process of an individual appearance texture driven by sequence human poses. In this article, we thereby propose to transform the synthesis into pose conditioned image generation task considering promising performance estimation technology and generative adversarial nets (GANs). However, problem is that GAN only suitable do region-aligned translation while involves large number spatial deformations. To avoid drawback, design two-step...

This paper presents an interactive image segmentation approach where the problem is formulated as a probabilistic estimation manner. Instead of measuring distances between unseeded pixels and seeded pixels, we measure similarities pixel pairs seed to improve robustness seeds. The unary prior probability each belonging foreground F background B can be effectively estimated based on with label (F, F),(F, B),(B, F) (B, B). Then likelihood learning framework proposed fuse region boundary...

Image and video based human motions can be regarded as the deformation processes of person appearances, so motion transfer is usually treated a pose guided image generation task implemented in 2D plane. However, plane lacks guidance original 3D information, which results blur shape distortions generated images. Therefore, we propose to simulate process real images by projecting models, are reconstructed from training driven with target poses, into We then take projections representations...