Digital Holography and Three-Dimensional Imaging, Journal Year: 2019, Volume and Issue: unknown, P. W2A.4 - W2A.4
Published: Jan. 1, 2019
We propose a holographic light field encoding and decoding technique, which can greatly reduce the data size. This technique has many potential applications, such as fast transfer, display for data.
Language: Английский
Optics Express, Journal Year: 2022, Volume and Issue: 30(13), P. 22590 - 22590
Published: April 27, 2022
Close-range 3D sensors based on the structured light principle have a constrained measuring range due to their depth of field (DOF). Focus stacking is method extend DOF. The additional time change focus drawback in high-speed measurements. In our research, chromatic was applied sensor with 180 fps frame rate. extended DOF evaluated by distance-dependent resolution derived from 3D-MTF tilted edge. conventional 14 mm 21 two foci at 455 and 520 nm wavelength. allowed shape measurements within 44 ms.
Language: Английский
Citations
6
Optics and Lasers in Engineering, Journal Year: 2019, Volume and Issue: 127, P. 105945 - 105945
Published: Nov. 18, 2019
Language: Английский
Citations
8
Published: Oct. 10, 2020
Microplastics, which are a major source of pollution in the ocean, need to be accurately detected and monitored. However, current detection approaches often require complex optical instrumentation long time for image processing. Furthermore, because difficulties particle sampling, it is hard collect dataset with sufficient images balanced distribution. Digital holography, non-destructive imaging method, suitable situ imaging. In this work, we propose novel digital holography microplastics classification system combines deep learning generative adversarial networks. We experimentally show that our method yields higher accuracy can efficiently reduce imbalance ratio dataset. This modified other tasks likewise suffer from small imbalanced distribution
Language: Английский
Citations
8
Optics Express, Journal Year: 2023, Volume and Issue: 31(12), P. 19766 - 19766
Published: May 15, 2023
Recently introduced, spaceplates achieve the propagation of light for a distance greater than their thickness. In this way, they compress optical space, reducing required between elements in an imaging system. Here we introduce spaceplate based on conventional optics 4-f arrangement, mimicking transfer function free-space thinner system - term device three-lens spaceplate. It is broadband, polarization-independent, and can be used meter-scale space compression. We experimentally measure compression ratios up to 15.6, replacing 4.4 meters free-space, three orders magnitude current spaceplates. demonstrate that reduce length full-color system, albeit with reductions resolution contrast. present theoretical limits numerical aperture ratio. Our design presents simple, accessible, cost-effective method optically compressing large amounts space.
Language: Английский
Citations
2
Published: Nov. 8, 2020
Light field (LF) imaging provides rich spatial and angular information, but is problematic in low-light environment as the images suffer from low contrast visibility. In this paper, we present a learning-based method to enhance LF images. A high-dimensional convolutional neural network (CNN) introduced extract spatio-angular features LF. The operates directly on four-dimensional data rather than individual sub-aperture images, avoiding loss of geometric information. Color compensation then performed enhanced coming CNN reduce color distortion. experimental results show that proposed achieves noticeable improvement compared with state-of-the-art image restoration techniques both visual inspection objective assessments.
Language: Английский
Citations
5
IEEE Transactions on Image Processing, Journal Year: 2021, Volume and Issue: 30, P. 3908 - 3921
Published: Jan. 1, 2021
This paper presents a learning-based approach to synthesize the view from an arbitrary camera position given sparse set of images. A key challenge for this novel synthesis arises reconstruction process, when views different input images may not be consistent due obstruction in light path. We overcome by jointly modeling epipolar property and occlusion designing convolutional neural network. start defining computing aperture disparity map, which approximates parallax measures pixel-wise shift between two views. While relates free-space rendering can fail near object boundaries, we further develop warping confidence map address pixel these challenging regions. The proposed method is evaluated on diverse real-world synthetic field scenes, it shows better performance over several state-of-the-art techniques.
Language: Английский
Citations
5
Sensors, Journal Year: 2021, Volume and Issue: 21(22), P. 7734 - 7734
Published: Nov. 20, 2021
Light field imaging plays an increasingly important role in the of three-dimensional (3D) reconstruction because its ability to quickly obtain four-dimensional information (angle and space) scene. In this paper, a 3D method light based on phase similarity is proposed increase accuracy depth estimation scope applicability epipolar plane image (EPI). The calibration camera was used relationship between disparity depth, projector removed make experimental procedure more flexible. Then, algorithm designed effectively improve reliability calculation, which instead structure tensor, morphological processing denoise optimize map. Finally, realized by combining with calibrated relationship. results showed that standard deviation two objects 0.3179 mm 0.3865 compared ground truth measured objects, respectively. Compared traditional EPI method, our can not only perform well single scene or blurred texture situations but also maintain good accuracy.
Language: Английский
Citations
5
Optics Express, Journal Year: 2022, Volume and Issue: 30(8), P. 13070 - 13070
Published: March 2, 2022
In this paper, we use deep neural networks (DNNs) to simultaneously reconstruct the amplitude and phase information of complex light field transmitted in atmospheric turbulence based on learning. The results reconstruction by four different training methods are compared comprehensively. obtained indicate that method can more accurately is input pattern pairs into network as two channels train model.
Language: Английский
Citations
3
Cureus, Journal Year: 2024, Volume and Issue: unknown
Published: May 17, 2024
Background: We developed a 3D camera system to track motion in surgical field. This has the potential introduce augmented reality (AR) systems non-invasively, eliminating need for invasive AR markers conventionally required. The present study was performed verify real-time tracking accuracy of this system, assess feasibility integrating into workflow, and establish its enhance efficiency orthopedic procedures. Methods: To evaluate technology using camera, forearm bone model created. depicted verified terms positional relationship with created from previously imaged CT data. Images field (capturing actual forearm) were taken saved nine poses by rotating pronation supination. alignment reference points computed at three versus yielding rotation matrix representing relationship. In original stereo vision-based depth image resolution 1280×720 pixels, 30 frames per second, lens view 64 specifications, baseline 3 cm, capable optimally acquiring data distance 40-60 cm subject used. modified following modifications made improve performance: (1) color filter processing changed HSV RGB, (2) detection supporting marker sizes 8 mm diameter, (3) positions stabilized calculating position each frame. Tracking examined parameters: differences matrix, maximum minimum inter-reference point errors between CT-based camera-based data, average error points. Results: difference matrices 5.51±2.68 mm. Average 1.10±0.61 15.53±12.51 mm, respectively. 6.26±4.49 4.22±1.73 0.79±0.49 1.94±0.87 1.41±0.58 once failed, it difficult recover accuracy. resulted large supination positions. These issues resolved system. Significant improvements achieved (P<0.05). Conclusion: developed. allows direct comparisons preoperative scans acquired camera. method advantage introducing without markers.
Language: Английский
Citations
0
Infrared and Laser Engineering, Journal Year: 2024, Volume and Issue: 53(9), P. 20240347 - 20240347
Published: Jan. 1, 2024
Citations
0