Noninvasive object imaging with single-shot low-resolution speckle pattern through strongly-scattering turbid layers

2020 ◽  
Author(s):  
Yingbo Wang ◽  
Jie Cao ◽  
Qun Hao ◽  
Chengqiang Xu ◽  
Mingyuan Tang
Keyword(s):  
Speckle Pattern ◽  
Single Shot ◽  
Low Resolution ◽  
Object Imaging

scholarly journals Small-Object Detection in Remote Sensing Images with End-to-End Edge-Enhanced GAN and Object Detector Network

2020 ◽  
Author(s):  
Jakaria Rabbi ◽  
Nilanjan Ray ◽  
Matthias Schubert ◽  
Subir Chowdhury ◽  
Dennis Chao
Keyword(s):  
Remote Sensing ◽  
Object Detection ◽  
Super Resolution ◽  
Detection Performance ◽  
Superior Performance ◽  
Single Shot ◽  
Small Object ◽  
Remote Sensing Images ◽  
Low Resolution ◽  
End To End

The detection performance of small objects in remote sensing images is not satisfactory compared to large objects, especially in low-resolution and noisy images. A generative adversarial network (GAN)-based model called enhanced super-resolution GAN (ESRGAN) shows remarkable image enhancement performance, but reconstructed images miss high-frequency edge information. Therefore, object detection performance degrades for the small objects on recovered noisy and low-resolution remote sensing images. Inspired by the success of edge enhanced GAN (EEGAN) and ESRGAN, we apply a new edge-enhanced super-resolution GAN (EESRGAN) to improve the image quality of remote sensing images and used different detector networks in an end-to-end manner where detector loss is backpropagated into the EESRGAN to improve the detection performance. We propose an architecture with three components: ESRGAN, Edge Enhancement Network (EEN), and Detection network. We use residual-in-residual dense blocks (RRDB) for both the GAN and EEN, and for the detector network, we use the faster region-based convolutional network (FRCNN) (two-stage detector) and single-shot multi-box detector (SSD) (one stage detector). Extensive experiments on car overhead with context and oil and gas storage tank (created by us) data sets show superior performance of our method compared to the standalone state-of-the-art object detectors.

scholarly journals Single-Shot Structured Light Sensor for 3D Dense and Dynamic Reconstruction

Sensors ◽  
2020 ◽  
Vol 20 (4) ◽  
pp. 1094 ◽  
Author(s):  
Feifei Gu ◽  
Zhan Song ◽  
Zilong Zhao
Keyword(s):  
Stereo Matching ◽  
Speckle Pattern ◽  
Structured Light ◽  
Optical Element ◽  
Depth Map ◽  
Point Clouds ◽  
Calibration Method ◽  
Acquisition Time ◽  
Single Shot ◽  
Dynamic Reconstruction

Structured light (SL) has a trade-off between acquisition time and spatial resolution. Temporally coded SL can produce a 3D reconstruction with high density, yet it is not applicable to dynamic reconstruction. On the contrary, spatially coded SL works with a single shot, but it can only achieve sparse reconstruction. This paper aims to achieve accurate 3D dense and dynamic reconstruction at the same time. A speckle-based SL sensor is presented, which consists of two cameras and a diffractive optical element (DOE) projector. The two cameras record images synchronously. First, a speckle pattern was elaborately designed and projected. Second, a high-accuracy calibration method was proposed to calibrate the system; meanwhile, the stereo images were accurately aligned by developing an optimized epipolar rectification algorithm. Then, an improved semi-global matching (SGM) algorithm was proposed to improve the correctness of the stereo matching, through which a high-quality depth map was achieved. Finally, dense point clouds could be recovered from the depth map. The DOE projector was designed with a size of 8 mm × 8 mm. The baseline between stereo cameras was controlled to be below 50 mm. Experimental results validated the effectiveness of the proposed algorithm. Compared with some other single-shot 3D systems, our system displayed a better performance. At close range, such as 0.4 m, our system could achieve submillimeter accuracy.

Single-Shot Colored Speckle Pattern for High Accuracy Depth Sensing

2019 ◽  
Vol 19 (17) ◽  
pp. 7591-7597 ◽  
Author(s):  
Boxun Fu ◽  
Fu Li ◽  
Tianjiao Zhang ◽  
Jingsong Jiang ◽  
Quanlu Li ◽  
...  
Keyword(s):  
Speckle Pattern ◽  
High Accuracy ◽  
Single Shot ◽  
Depth Sensing

scholarly journals Small-Object Detection in Remote Sensing Images with End-to-End Edge-Enhanced GAN and Object Detector Network

2020 ◽  
Author(s):  
Jakaria Rabbi ◽  
Nilanjan Ray ◽  
Matthias Schubert ◽  
Subir Chowdhury ◽  
Dennis Chao
Keyword(s):  
Remote Sensing ◽  
Object Detection ◽  
Super Resolution ◽  
Detection Performance ◽  
Superior Performance ◽  
Single Shot ◽  
Small Object ◽  
Remote Sensing Images ◽  
Low Resolution ◽  
End To End

The detection performance of small objects in remote sensing images is not satisfactory compared to large objects, especially in low-resolution and noisy images. A generative adversarial network (GAN)-based model called enhanced super-resolution GAN (ESRGAN) shows remarkable image enhancement performance, but reconstructed images miss high-frequency edge information. Therefore, object detection performance degrades for small objects on recovered noisy and low-resolution remote sensing images. Inspired by the success of edge enhanced GAN (EEGAN) and ESRGAN, we apply a new edge-enhanced super-resolution GAN (EESRGAN) to improve the image quality of remote sensing images and use different detector networks in an end-to-end manner where detector loss is backpropagated into the EESRGAN to improve the detection performance. We propose an architecture with three components: ESRGAN, Edge Enhancement Network (EEN), and Detection network. We use residual-in-residual dense blocks (RRDB) for both the ESRGAN and EEN, and for the detector network, we use the faster region-based convolutional network (FRCNN) (two-stage detector) and single-shot multi-box detector (SSD) (one stage detector). Extensive experiments on a public (car overhead with context) and a self-assembled (oil and gas storage tank) satellite dataset show superior performance of our method compared to the standalone state-of-the-art object detectors.

scholarly journals Residual Super-Resolution Single Shot Network for Low-Resolution Object Detection

IEEE Access ◽  
2018 ◽  
Vol 6 ◽  
pp. 47780-47793 ◽  
Author(s):  
Xiaotong Zhao ◽  
Wei Li ◽  
Yifan Zhang ◽  
Zhiyong Feng
Keyword(s):  
Object Detection ◽  
Super Resolution ◽  
Single Shot ◽  
Low Resolution ◽  
Resolution Single

scholarly journals An Efficient Low-Resolution NIR Classification Scheme for M, L, and T dwarfs and Its Application to Young BDs

2003 ◽  
Vol 211 ◽  
pp. 359-360
Author(s):  
L. Testi ◽  
A. Natta ◽  
C. Baffa ◽  
G. Comoretto ◽  
S. Gennari ◽  
...  
Keyword(s):  
Near Infrared ◽  
Classification Scheme ◽  
Optical Element ◽  
System Throughput ◽  
Single Shot ◽  
Low Resolution ◽  
Near Infrared Spectra ◽  
Infrared Spectral ◽  
High System ◽  
Large Telescopes

We present the preliminary results of a programme aimed at defining a low-resolution near-infrared spectral classification scheme for faint M, L, and T-dwarfs. The method is based on the global shape of R˜100 complete near-infrared spectra from 0.8 to 2.4μm as obtained through a high-throughput prism-based optical element, the Amici device, mounted inside the NICS instrument at the TNG 3.5m telescope (Baffa et al. 2001; Oliva 2000). The aim of our project is to provide an efficient classification scheme based on very low-resolution near infrared spectroscopy, which can be carried on at a 4m-class telescope. The results for the L-type dwarfs have already been presented in Testi et al. (2001), sample spectra for the M and T-dwarfs range are shown in Figure 1. A preliminary application of the method to the classification of young embedded brown-dwarf candidates has been successfully attempted by Testi et al. (2002) and Natta et al. (2002). The method is shown to be accurate and competitive: the high system throughput coupled with the possibility of obtaining in a “single shot” the complete spectrum of the objects make the NICS/TNG system more efficient than existing large telescopes.

scholarly journals Small-Object Detection in Remote Sensing Images with End-to-End Edge-Enhanced GAN and Object Detector Network

2020 ◽  
Author(s):  
Jakaria Rabbi ◽  
Nilanjan Ray ◽  
Matthias Schubert ◽  
Subir Chowdhury ◽  
Dennis Chao
Keyword(s):  
Remote Sensing ◽  
Object Detection ◽  
Super Resolution ◽  
Detection Performance ◽  
Superior Performance ◽  
Single Shot ◽  
Small Object ◽  
Remote Sensing Images ◽  
Low Resolution ◽  
End To End

The detection performance of small objects in remote sensing images has not been satisfactory compared to large objects, especially in low-resolution and noisy images. A generative adversarial network (GAN)-based model called enhanced super-resolution GAN (ESRGAN) showed remarkable image enhancement performance, but reconstructed images usually miss high-frequency edge information. Therefore, object detection performance showed degradation for small objects on recovered noisy and low-resolution remote sensing images. Inspired by the success of edge enhanced GAN (EEGAN) and ESRGAN, we applied a new edge-enhanced super-resolution GAN (EESRGAN) to improve the quality of remote sensing images and used different detector networks in an end-to-end manner where detector loss was backpropagated into the EESRGAN to improve the detection performance. We proposed an architecture with three components: ESRGAN, EEN, and Detection network. We used residual-in-residual dense blocks (RRDB) for both the ESRGAN and EEN, and for the detector network, we used a faster region-based convolutional network (FRCNN) (two-stage detector) and a single-shot multibox detector (SSD) (one stage detector). Extensive experiments on a public (car overhead with context) dataset and another self-assembled (oil and gas storage tank) satellite dataset showed superior performance of our method compared to the standalone state-of-the-art object detectors.

Sign in / Sign up

Export Citation Format

Share Document