Improved SfM-Based Indoor Localization with Occlusion Removal

This article describes a novel 3D image-based indoor localization system integrated with an improved SfM (structure from motion) approach and an obstacle removal component. In contrast with existing state-of-the-art localization techniques focusing on static outdoor or indoor environments, the adverse effects, generated by moving obstacles in busy indoor spaces, are considered in this work. In particular, the problem of occlusion removal is converted into a separation problem of moving foreground and static background. A low-rank and sparse matrix decomposition approach is used to solve this problem efficiently. Moreover, a SfM with RT (re-triangulation) is adopted in order to handle the drifting problem of incremental SfM method in indoor scene reconstruction. To evaluate the performance of the system, three data sets and the corresponding query sets are established to simulate different states of the indoor environment. Quantitative experimental results demonstrate that both query registration rate and localization accuracy increase significantly after integrating the authors' improvements.

Download Full-text

Musical noise suppression using a low-rank and sparse matrix decomposition approach

Speech Communication ◽

10.1016/j.specom.2020.09.001 ◽

2020 ◽

Vol 125 ◽

pp. 41-52

Author(s):

Jishnu Sadasivan ◽

Jitendra K. Dhiman ◽

Chandra Sekhar Seelamantula

Keyword(s):

Noise Suppression ◽

Sparse Matrix ◽

Matrix Decomposition ◽

Low Rank ◽

Decomposition Approach

Download Full-text

Efficient background subtraction with low-rank and sparse matrix decomposition

2015 IEEE International Conference on Image Processing (ICIP) ◽

10.1109/icip.2015.7351731 ◽

2015 ◽

Cited By ~ 13

Author(s):

Salehe Erfanian Ebadi ◽

Valia Guerra Ones ◽

Ebroul Izquierdo

Keyword(s):

Background Subtraction ◽

Sparse Matrix ◽

Matrix Decomposition ◽

Low Rank

Download Full-text

Orthogonal Subspace Projection Using Data Sphering and Low-Rank and Sparse Matrix Decomposition for Hyperspectral Target Detection

IEEE Transactions on Geoscience and Remote Sensing ◽

10.1109/tgrs.2021.3053201 ◽

2021 ◽

pp. 1-19

Author(s):

Chein-I Chang ◽

Jie Chen

Keyword(s):

Target Detection ◽

Sparse Matrix ◽

Matrix Decomposition ◽

Low Rank ◽

Subspace Projection ◽

Orthogonal Subspace ◽

Orthogonal Subspace Projection ◽

Using Data

Download Full-text

Low-Rank and Sparse Matrix Decomposition With Orthogonal Subspace Projection-Based Background Suppression for Hyperspectral Anomaly Detection

IEEE Geoscience and Remote Sensing Letters ◽

10.1109/lgrs.2019.2948675 ◽

2020 ◽

Vol 17 (8) ◽

pp. 1378-1382

Author(s):

Yixin Yang ◽

Jianqi Zhang ◽

Shangzhen Song ◽

Chi Zhang ◽

Delian Liu

Keyword(s):

Anomaly Detection ◽

Sparse Matrix ◽

Matrix Decomposition ◽

Low Rank ◽

Background Suppression ◽

Subspace Projection ◽

Orthogonal Subspace ◽

Orthogonal Subspace Projection

Download Full-text

A Visual and VAE Based Hierarchical Indoor Localization Method

Sensors ◽

10.3390/s21103406 ◽

2021 ◽

Vol 21 (10) ◽

pp. 3406

Author(s):

Jie Jiang ◽

Yin Zou ◽

Lidong Chen ◽

Yujie Fang

Keyword(s):

Image Retrieval ◽

Indoor Localization ◽

Data Sets ◽

Indoor Environments ◽

Global Features ◽

Data Set ◽

Data Annotation ◽

Wide Range ◽

Annotation Costs ◽

Global And Local

Precise localization and pose estimation in indoor environments are commonly employed in a wide range of applications, including robotics, augmented reality, and navigation and positioning services. Such applications can be solved via visual-based localization using a pre-built 3D model. The increase in searching space associated with large scenes can be overcome by retrieving images in advance and subsequently estimating the pose. The majority of current deep learning-based image retrieval methods require labeled data, which increase data annotation costs and complicate the acquisition of data. In this paper, we propose an unsupervised hierarchical indoor localization framework that integrates an unsupervised network variational autoencoder (VAE) with a visual-based Structure-from-Motion (SfM) approach in order to extract global and local features. During the localization process, global features are applied for the image retrieval at the level of the scene map in order to obtain candidate images, and are subsequently used to estimate the pose from 2D-3D matches between query and candidate images. RGB images only are used as the input of the proposed localization system, which is both convenient and challenging. Experimental results reveal that the proposed method can localize images within 0.16 m and 4° in the 7-Scenes data sets and 32.8% within 5 m and 20° in the Baidu data set. Furthermore, our proposed method achieves a higher precision compared to advanced methods.

Download Full-text