Extracting image features in static images for depth estimation

2013 ◽  
Author(s):  
Masaki Ogino ◽  
Junji Suzuki ◽  
Minoru Asada
Keyword(s):  
Depth Estimation ◽  
Image Features ◽  
Static Images

scholarly journals Latent 3D Volume for Joint Depth Estimation and Semantic Segmentation from a Single Image

Sensors ◽  
2020 ◽  
Vol 20 (20) ◽  
pp. 5765 ◽  
Author(s):  
Seiya Ito ◽  
Naoshi Kaneko ◽  
Kazuhiko Sumi
Keyword(s):  
3D Structure ◽  
Three Dimensional ◽  
Semantic Segmentation ◽  
Depth Estimation ◽  
Image Features ◽  
Feature Representation ◽  
Single Image ◽  
Feature Vectors ◽  
3D Space ◽  
3D Volume

This paper proposes a novel 3D representation, namely, a latent 3D volume, for joint depth estimation and semantic segmentation. Most previous studies encoded an input scene (typically given as a 2D image) into a set of feature vectors arranged over a 2D plane. However, considering the real world is three-dimensional, this 2D arrangement reduces one dimension and may limit the capacity of feature representation. In contrast, we examine the idea of arranging the feature vectors in 3D space rather than in a 2D plane. We refer to this 3D volumetric arrangement as a latent 3D volume. We will show that the latent 3D volume is beneficial to the tasks of depth estimation and semantic segmentation because these tasks require an understanding of the 3D structure of the scene. Our network first constructs an initial 3D volume using image features and then generates latent 3D volume by passing the initial 3D volume through several 3D convolutional layers. We apply depth regression and semantic segmentation by projecting the latent 3D volume onto a 2D plane. The evaluation results show that our method outperforms previous approaches on the NYU Depth v2 dataset.

Dynamic IBVS of a rotary wing UAV using line features

Robotica ◽  
2014 ◽  
Vol 34 (9) ◽  
pp. 2009-2026 ◽  
Author(s):  
Hui Xie ◽  
Alan F. Lynch ◽  
Martin Jagersand
Keyword(s):  
Visual Servoing ◽  
Depth Estimation ◽  
Image Features ◽  
Interaction Matrix ◽  
Globally Asymptotically Stable ◽  
Virtual Camera ◽  
Control Objective ◽  
Aerial Vehicle ◽  
Inspection Tasks ◽  
Rotary Wing

SUMMARYIn this paper we propose a dynamic image-based visual servoing (IBVS) control for a rotary wing unmanned aerial vehicle (UAV) which directly accounts for the vehicle's underactuated dynamic model. The motion control objective is to follow parallel lines and is motivated by power line inspection tasks where the UAV's relative position and orientation to the lines are controlled. The design is based on a virtual camera whose motion follows the onboard physical camera but which is constrained to point downwards independent of the vehicle's roll and pitch angles. A set of image features is proposed for the lines projected into the virtual camera frame. These features are chosen to simplify the interaction matrix which in turn leads to a simpler IBVS control design which is globally asymptotically stable. The proposed scheme is adaptive and therefore does not require depth estimation. Simulation results are presented to illustrate the performance of the proposed control and its robustness to calibration parameter error.

scholarly journals Structure-Aware Residual Pyramid Network for Monocular Depth Estimation

2019 ◽  
Author(s):  
Xiaotian Chen ◽  
Xuejin Chen ◽  
Zheng-Jun Zha
Keyword(s):  
Local Structure ◽  
Feature Fusion ◽  
Depth Estimation ◽  
Image Features ◽  
Underlying Structure ◽  
Complex Scene ◽  
Multi Scale ◽  
Depth Prediction ◽  
Scale Structures ◽  
Monocular Depth

Monocular depth estimation is an essential task for scene understanding. The underlying structure of objects and stuff in a complex scene is critical to recovering accurate and visually-pleasing depth maps. Global structure conveys scene layouts, while local structure reflects shape details. Recently developed approaches based on convolutional neural networks (CNNs) significantly improve the performance of depth estimation. However, few of them take into account multi-scale structures in complex scenes. In this paper, we propose a Structure-Aware Residual Pyramid Network (SARPN) to exploit multi-scale structures for accurate depth prediction. We propose a Residual Pyramid Decoder (RPD) which expresses global scene structure in upper levels to represent layouts, and local structure in lower levels to present shape details. At each level, we propose Residual Refinement Modules (RRM) that predict residual maps to progressively add finer structures on the coarser structure predicted at the upper level. In order to fully exploit multi-scale image features, an Adaptive Dense Feature Fusion (ADFF) module, which adaptively fuses effective features from all scales for inferring structures of each scale, is introduced. Experiment results on the challenging NYU-Depth v2 dataset demonstrate that our proposed approach achieves state-of-the-art performance in both qualitative and quantitative evaluation. The code is available at https://github.com/Xt-Chen/SARPN.

Structured Light Field Generated by Two Projectors for High-Speed Three Dimensional Measurement

2016 ◽  
Vol 28 (4) ◽  
pp. 523-532 ◽  
Author(s):  
Akihiro Obara ◽  
◽  
Xu Yang ◽  
Hiromasa Oku ◽  
Keyword(s):  
High Speed ◽  
Light Field ◽  
Stereo Matching ◽  
Structured Light ◽  
Three Dimensional ◽  
Depth Estimation ◽  
Image Features ◽  
Depth Information ◽  
Speed Tracking ◽  
Three Dimensional Measurement

[abstFig src='/00280004/10.jpg' width='300' text='Concept of SLF generated by two projectors' ] Triangulation is commonly used to restore 3D scenes, but its frame of less than 30 fps due to time-consuming stereo-matching is an obstacle for applications requiring that results be fed back in real time. The structured light field (SLF) our group proposed previously reduced the amount of calculation in 3D restoration, realizing high-speed measurement. Specifically, the SLF estimates depth information by projecting information on distance directly to a target. The SLF synthesized as reported, however, presents difficulty in extracting image features for depth estimation. In this paper, we propose synthesizing the SLF using two projectors with a certain layout. Our proposed SLF’s basic properties are based on an optical model. We evaluated the SLF’s performance using a prototype we developed and applied to the high-speed depth estimation of a target moving randomly at a speed of 1000 Hz. We demonstrate the target’s high-speed tracking based on high-speed depth information feedback.

scholarly journals Unsupervised monocular depth estimation with aggregating image features and wavelet SSIM (Structural SIMilarity) loss

2021 ◽  
Author(s):  
Bingen Li ◽  
Hao Zhang ◽  
Zhuping Wang ◽  
Chun Liu ◽  
Huaicheng Yan ◽  
...  
Keyword(s):  
Structural Similarity ◽  
Depth Estimation ◽  
Image Features ◽  
Monocular Depth

7-beam lattice images of (110) oriented Ge

1978 ◽  
Vol 36 (1) ◽  
pp. 290-291
Author(s):  
J.R. Parsons ◽  
C.W. Hoelke
Keyword(s):  
Image Features ◽  
Objective Lens ◽  
Lattice Plane ◽  
Lattice Image ◽  
Crystalline Defects ◽  
Transmission Electron ◽  
Lattice Images ◽  
Electron Microscopes ◽  
Structural Aspects ◽  
Beam Lattice

The direct imaging of a crystal lattice has intrigued electron microscopists for many years. What is of interest, of course, is the way in which defects perturb their atomic regularity. There are problems, however, when one wishes to relate aperiodic image features to structural aspects of crystalline defects. If the defect is inclined to the foil plane and if, as is the case with present 100 kV transmission electron microscopes, the objective lens is not perfect, then terminating fringes and fringe bending seen in the image cannot be related in a simple way to lattice plane geometry in the specimen (1).The purpose of the present work was to devise an experimental test which could be used to confirm, or not, the existence of a one-to-one correspondence between lattice image and specimen structure over the desired range of specimen spacings. Through a study of computed images the following test emerged.

High-resolution electron microscopy of the atomic structure of some grain boundaries in Au

1986 ◽  
Vol 44 ◽  
pp. 414-415
Author(s):  
W. Krakow ◽  
D. A. Smith
Keyword(s):  
Atomic Structure ◽  
High Resolution Electron Microscopy ◽  
Image Features ◽  
Image Feature ◽  
Resolution Electron ◽  
Tilt Boundaries ◽  
Processing Techniques ◽  
Atom Position ◽  
Structure Configuration

The successful determination of the atomic structure of [110] tilt boundaries in Au stems from the investigation of microscope performance at intermediate accelerating voltages (200 and 400kV) as well as a detailed understanding of how grain boundary image features depend on dynamical diffraction processes variation with specimen and beam orientations. This success is also facilitated by improving image quality by digital image processing techniques to the point where a structure image is obtained and each atom position is represented by a resolved image feature. Figure 1 shows an example of a low angle (∼10°) Σ = 129/[110] tilt boundary in a ∼250Å Au film, taken under tilted beam brightfield imaging conditions, to illustrate the steps necessary to obtain the atomic structure configuration from the image. The original image of Fig. 1a shows the regular arrangement of strain-field images associated with the cores of ½ [10] primary dislocations which are separated by ∼15Å.

Imaging of polymer single crystals in low-voltage, high-resolution scanning electron microscopy

1990 ◽  
Vol 48 (4) ◽  
pp. 1106-1107
Author(s):  
W.W. Adams ◽  
G. Price ◽  
A. Krause
Keyword(s):  
Electron Microscopy ◽  
Scanning Electron Microscopy ◽  
High Resolution ◽  
Single Crystals ◽  
Low Voltage ◽  
Polymer Surface ◽  
Beam Current ◽  
Image Features ◽  
First Results ◽  
Scanning Electron

It has been shown that there are numerous advantages in imaging both coated and uncoated polymers in scanning electron microscopy (SEM) at low voltages (LV) from 0.5 to 2.0 keV compared to imaging at conventional voltages of 10 to 20 keV. The disadvantages of LVSEM of degraded resolution and decreased beam current have been overcome with the new generation of field emission gun SEMs. In imaging metal coated polymers in LVSEM beam damage is reduced, contrast is improved, and charging from irregularly shaped features (which may be unevenly coated) is reduced or eliminated. Imaging uncoated polymers in LVSEM allows direct observation of the surface with little or no charging and with no alterations of surface features from the metal coating process required for higher voltage imaging. This is particularly important for high resolution (HR) studies of polymers where it is desired to image features 1 to 10 nm in size. Metal sputter coating techniques produce a 10 - 20 nm film that has its own texture which can obscure topographical features of the original polymer surface. In examining thin, uncoated insulating samples on a conducting substrate at low voltages the effect of sample-beam interactions on image formation and resolution will differ significantly from the effect at higher accelerating voltages. We discuss here sample-beam interactions in single crystals on conducting substrates at low voltages and also present the first results on HRSEM of single crystal morphologies which show some of these effects.

Sign in / Sign up

Export Citation Format

Share Document