Computational Ghost Imaging (CGI)
Generation of Speckles via Deep Learning
Detailed description can be found here
In this project, we present a method for speckle pattern design using deep learning. The speckle patterns possess unique features after experiencing convolutions in Speckle-Net, our well-designed framework for speckle pattern generation. We then apply our method to the computational ghost imaging system. The standard deep learning-assisted ghost imaging methods use the network to recognize the reconstructed objects or imaging algorithms. In contrast, this innovative application optimizes the illuminating speckle patterns via Speckle-Net with specific sampling rates. Our method, therefore, outperforms the other techniques for ghost imaging, particularly its ability to retrieve high-quality images with extremely low sampling rates. It opens a new route towards non-trivial speckle generation by referring to a standard loss function on specified objectives with the modified deep neural network. It also has great potential in other areas using speckle patterns such as dynamic speckle illumination microscopy, structured illumination microscopy, x-ray imaging, photo-acoustic imaging, and optical lattices.
0.5% Nyquist Imaging via Deep Learning (Peer review)
arXiv:2108.07673
We present a novel framework for computational ghost imaging based on deep learning and pink noise patterns, which substantially decreases the sampling ratio over 10 times than previous sub-Nyquist computational ghost imaging works. Here, the deep neural network, which can learn the sensing model and enhance image reconstruction quality, is trained merely by simulation results. There is no necessity to conduct experiments to get training inputs (non-experimental) and add noise to customize with a real imaging system (noise-free). This one-time trained network can be applied to multiple environments and various situations. To demonstrate the sub-Nyquist level in our achievement, the conventional computational ghost imaging results, imaging results reconstructed using white noise, and pink noise via deep learning are compared in several sampling rates. To indicate its non-experimental and noise-free advantages, a group of results with strong environmental noise is presented. This method has great potentials in various applications that require a low sampling rate, quick reconstruction efficiency, and strong turbulence.