- 标题
- 摘要
- 关键词
- 实验方案
- 产品
过滤筛选
- 2018
- railway track
- kinematic characteristics
- stereo vision
- image blur
- distance estimation
- control system
- depth of field
- video monitoring
- hollows classification
- RGB-D camera
- Optoelectronic Information Science and Engineering
- Chongqing University of Posts and Telecommunications
- Bauman Moscow State Technical University
- Moscow State University of Railway Engineering
-
DeepLens
摘要： We aim to generate high resolution shallow depth-of-field (DoF) images from a single all-in-focus image with controllable focal distance and aperture size. To achieve this, we propose a novel neural network model comprised of a depth prediction module, a lens blur module, and a guided upsampling module. All modules are differentiable and are learned from data. To train our depth prediction module, we collect a dataset of 2462 RGB-D images captured by mobile phones with a dual-lens camera, and use existing segmentation datasets to improve border prediction. We further leverage a synthetic dataset with known depth to supervise the lens blur and guided upsampling modules. The effectiveness of our system and training strategies are verified in the experiments. Our method can generate high-quality shallow DoF images at high resolution, and produces significantly fewer artifacts than the baselines and existing solutions for single image shallow DoF synthesis. Compared with the iPhone portrait mode, which is a state-of-the-art shallow DoF solution based on a dual-lens depth camera, our method generates comparable results, while allowing for greater flexibility to choose focal points and aperture size, and is not limited to one capture setup.
关键词： Neural Network,Shallow Depth of Field
更新于2025-09-23 15:23:52
-
[ACM Press the Ninth International Symposium - Danang City, Viet Nam (2018.12.06-2018.12.07)] Proceedings of the Ninth International Symposium on Information and Communication Technology - SoICT 2018 - Calf Robust Weight Estimation Using 3D Contiguous Cylindrical Model and Directional Orientation from Stereo Images
摘要： Calving interval is often used as an indicator for fertility of beef cattle, however, maternal abilities are also required because the value of breeding cows depends on how efficiently the healthy and growing calves are produced. The calf's weight has been used as an indicator of maternity ability since the past few decades. We propose a method to estimate body weight by modeling the shape of calf using 3D information extracted from the stereo images. This method enables to predict the swelling of the cattle's body by creating a 3D model, which cannot be obtained solely from a 2D image. In addition, it is possible to estimate robust weight regardless of different shooting conditions toward cattle's posture and orientation. An image suitable for estimation is selected from motion images taken by the camera installed in the barn, and 3D coordinates are calculated by the images. Then, only the body is developed with a 3D model as it has the highest correlation with the body weight. Considering that the side of cattle's body may not be exactly perpendicular to the camera's shooting direction, a symmetric axis is extracted to find the inclination of cattle body from the camera in order to generate a 3D model based on the symmetric axis. 3D contiguous cylindrical model is used for the body of a cattle which has a rounded shape. In order to manipulate the shapes of the cylindrical surface, the circle and ellipse fittings are applied and compared. The linear regression equation of the volume of the cylindrical model and the actually measured body weight are used to estimate the cattle weight. As a result of modeling with the proposed method using the actual camera images, the correlation coefficient between the body weight and the model volume was at the best value, 0.9107. Even when experimentally examined with the different 3D coordinates obtained from other types of camera, the MAPE (Mean Absolute Percentage Error) was as low as 6.39%.
关键词： depth camera,circle fitting,three-dimensional reconstruction,weight estimation,calf,cow,stereo camera,cattle,ellipse fitting
更新于2025-09-23 15:23:52
-
Climatological analysis of the optical properties of aerosols and their direct radiative forcing in the Middle East
摘要： In addition to climate perturbations, various problems such as air pollution, reduction in the visibility and human health hazards were caused by atmospheric aerosols in the Middle East specifically in the last two decades. With the help of the Aerosol Robotic NETwork (AERONET), the measurement of the aerosol optical and radiative properties were carried out over seven sites in the Middle East during 2013. The analysis of the optical properties of aerosols like Single Scattering Albedo (SSA), Angstrom Exponent (AE), Aerosol Optical Depth (AOD), and Asymmetry parameter (ASY) were carried out during the study period. During spring and summer, high values of AOD and low values of AE were found in all sites except CUT-TEPAK (Limassol, Cyprus), which specified the existence of coarse mode particles and dust storms in these seasons. The AE maximum values were found in the summer and fall over CUT-TEPAK and IMS-METU-ERDEMLI(Erdemli, Turkey), whereas in other sites IASBS (Zanjan, Iran), KAUST Campus (Thuwal, Saudi Arabia), Masdar Institute (Masdar, United Arab Emirates), Mezaira (Mezaira, United Arab Emirates) and Solar Village (Riyadh, Saudi Arabia) the peak values of AE occurred in the fall and winter. The maximum values of SSA and ASY were observed in the spring and summer over all sites except over CUT-TEPAK and IMS-METU-ERDEMLI. The Santa Barbara DISORT Atmospheric Radiative Transfer (SBDART) model has been used for the calculations of the Aerosol Radiative Forcing (ARF) over the selected sites. We obtained negative value of ARF at the surface, which suggesting its cooling effects because of the loss of radiation back to space due to aerosols. The averaged ARF values at the SuRFace (SRF) of the earth were -43.8 Wm-2, -31 Wm-2, -56.8 Wm-2, -61.7 Wm-2, -52.5 Wm-2, -54.9 Wm-2, and -72.2 Wm-2, over CUT-TEPAK, IASABS, IMS-METU-ERDEMLI, KAUST Campus, Masdar Institute, Mezaira and Solar Village, respectively. While the positive value of atmospheric ARF showed heating of the atmosphere.
关键词： Middle East.,Aerosol Optical Depth,SBDART,Aerosol Radiative Forcing,AERONET
更新于2025-09-23 15:23:52
-
[ACM Press SIGGRAPH Asia 2018 Posters - Tokyo, Japan (2018.12.04-2018.12.07)] SIGGRAPH Asia 2018 Posters on - SA '18 - Panoramic depth reconstruction within a single shot by optimizing global sphere radii
摘要： Depth estimation in scene reconstruction remains one of the main issues in the world of virtual reality. We propose a method that uses low cost camera facilities from previous papers and their specified procedures for stereoscopic 360 imaging. However, instead of using angular disparity, we use the spherical radius for labeling depth values for scene reconstruction. The experimental results show that the reconstructed shape is less distorted by directly optimizing spherical radii than optimizing the angular disparities.
关键词： Matching cost map,Polycamera,Inverse radius disparity depth estimation,Depth Reconstruction,Depth map
更新于2025-09-23 15:22:29
-
Vision-based people detection using depth information for social robots
摘要： Robots are starting to be applied in areas which involve sharing space with humans. In particular, social robots and people will coexist closely because the former are intended to interact with the latter. In this context, it is crucial that robots are aware of the presence of people around them. Traditionally, people detection has been performed using a flow of two-dimensional images. However, in nature, animals’ sight perceives their surroundings using color and depth information. In this work, we present new people detectors that make use of the data provided by depth sensors and red-green-blue images to deal with the characteristics of human–robot interaction scenarios. These people detectors are based on previous works using two-dimensional images and existing people detectors from different areas. The disparity of the input and output data used by these types of algorithms usually complicates their integration into robot control architectures. We propose a common interface that can be used by any people detector, resulting in numerous advantages. Several people detectors using depth information and the common interface have been implemented and evaluated. The results show a great diversity among the different algorithms. Each one has a particular domain of use, which is reflected in the results. A clever combination of several algorithms appears as a promising solution to achieve a flexible, reliable people detector.
关键词： kinect,people detection,benchmarking,depth image processing,depth sensor,Social robotics,user detection,ROS
更新于2025-09-23 15:22:29
-
A Novel Design of Through-Hole Depth On-Machine Optical Measuring Equipment for Automatic Drilling and Riveting
摘要： In the aerospace manufacturing industry, it is impossible to achieve precise and efficient automatic drilling and riveting for largescale composite board parts. The bottleneck is that the depth detection of rivet holes still relies on manual operation, which seriously affects the assembly efficiency and stability of composite board parts. In order to realize accurate and efficient on-machine automatic measurement for through holes in the automatic drilling and riveting process of largescale composite board parts, this paper presents a novel hole depth measuring device. Its mechanical structure is developed based on our newly designed measurement scheme and optical path, the purpose of which is to convert the hole depth data into displacement data of the probe motion. Its electrical hardware consists of three units: a laser transceiver unit to pick up laser spots; a displacement measuring unit to capture the probe movement in real time; and a driving unit to achieve motion control of the probe. Finally, the experimental results indicated that the proposed method and device are capable of performing automatic measurements for through-hole depth. In addition, factors affecting the measuring accuracy and stability of the device are initially analyzed and discussed, which lay a foundation for subsequent research on error compensation and probe calibration.
关键词： image processing,depth detection,automatic drilling and riveting,through-hole depth,large-scale composite board,on-machine measurement
更新于2025-09-23 15:22:29
-
[IEEE 2018 14th International Conference on Emerging Technologies (ICET) - Islamabad, Pakistan (2018.11.21-2018.11.22)] 2018 14th International Conference on Emerging Technologies (ICET) - Identification and mapping of coral reefs using Landsat 8 OLI in Astola Island, Pakistan coastal ocean
摘要： Recent field surveys have reported the presence of corals in many places in the Pakistan coastal ocean; Astola Island especially has been a subject of interest with regards to corals and overall marine biodiversity, and has in fact recently been declared Pakistan’s first Marine Protected Area. This study presents an analysis of coral reefs identification and their spatial distribution through optical satellite remote sensing in the surrounding area of Astola Island. Besides remote sensing data, the study considers sea survey data collected by divers in recent years. A benthic map of ocean ecosystem habitats is generated, through processing of Landsat 8 OLI (Operational Land Imager) imagery. The satellite data was selected at low-tide time to get maximum sunlight penetration in shallow water. Water-column correction was used to generate the depth-invariant index on multiple band-pairs. Water column corrected depth-invariant index bands were then segmented and classified through object-based classification. The results from the remote sensing data processing over Astola Island show good agreement with the field survey data, with nearly all the field survey points of coral reefs falling within the coral reefs class. Use of remote sensing imagery such as Landsat 8, and application of the water column correction method can allow for regular monitoring and management of coral reefs and other benthic ecosystems in the coastal ocean of Pakistan and coastal Arabian Sea.
关键词： Pan-sharpening,Landsat 8,Coral reefs,Object Base Image Analysis,Depth Invariant Index
更新于2025-09-23 15:22:29
-
[IEEE 2018 25th IEEE International Conference on Image Processing (ICIP) - Athens, Greece (2018.10.7-2018.10.10)] 2018 25th IEEE International Conference on Image Processing (ICIP) - Stereo Generation from a Single Image Using Deep Residual Network
摘要： In this paper, we propose a framework to generate stereoscopic content from a single image using the relative depth label predicted from deep residual network. Specifically, our framework first obtains a coarse relative depth label from the network and refines it to painting depth by sampling and interpolation, then an unsupervised clustering algorithm is employed to separate pixels of different depths into different layers to generate stereoscopic images. Experimental results with good visual effects demonstrate that the proposed method can be generally applied in both outdoor and indoor scenes. Meanwhile the quantitative results on relative depth estimation from a single image are comparable to state-of-the-art. Further experiments show the application possibility of our method in VR and panorama.
关键词： layered images,residual networks,relative depth,Stereo generation
更新于2025-09-23 15:22:29
-
[IEEE 2018 5th IEEE International Workshop on Metrology for AeroSpace (MetroAeroSpace) - Rome, Italy (2018.6.20-2018.6.22)] 2018 5th IEEE International Workshop on Metrology for AeroSpace (MetroAeroSpace) - Characterization and Testing of a High-Resolution Time-of-Flight Camera for Autonomous Navigation
摘要： This paper presents the results of research activities carried out to characterize and test the operation of a latest generation commercial, high-resolution Time-of-Flight (TOF) camera. The aim is to preliminary evaluate the achievable performance as well as potential limitations related to the use of this instrument for autonomous navigation purposes. Two fields of investigation have been identified: autonomous navigation of Unmanned Aerial Vehicles flying in GPS-denied environments; autonomous relative navigation between non-cooperative space objects. With reference to these applications, first, a metrological characterization has been operated within a laboratory setup. Second, experimental tests have been carried out by processing point clouds acquired by the TOF sensor with state-of-the-art algorithms (for depth-based odometry and non-cooperative pose determination, respectively).
关键词： depth-based odometry,non-cooperative pose determination,TOF camera,autonomous navigation,metrological characterization
更新于2025-09-23 15:22:29
-
Multiview Layer Fusion Model for Action Recognition Using RGBD Images
摘要： Vision-based action recognition encounters different challenges in practice, including recognition of the subject from any viewpoint, processing of data in real time, and offering privacy in a real-world setting. Even recognizing profile-based human actions, a subset of vision-based action recognition, is a considerable challenge in computer vision which forms the basis for an understanding of complex actions, activities, and behaviors, especially in healthcare applications and video surveillance systems. Accordingly, we introduce a novel method to construct a layer feature model for a profile-based solution that allows the fusion of features for multiview depth images. This model enables recognition from several viewpoints with low complexity at a real-time running speed of 63 fps for four profile-based actions: standing/walking, sitting, stooping, and lying. The experiment using the Northwestern-UCLA 3D dataset resulted in an average precision of 86.40%. With the i3DPost dataset, the experiment achieved an average precision of 93.00%. With the PSU multiview profile-based action dataset, a new dataset for multiple viewpoints which provides profile-based action RGBD images built by our group, we achieved an average precision of 99.31%.
关键词： privacy-preserving surveillance,layer fusion model,real-time processing,Multiview action recognition,RGBD images,depth-based features
更新于2025-09-23 15:22:29