Recently, free-space optical neural networks (ONNs) have gained extensive interest as emerging machine learning platforms for implementing artificial intelligence tasks, such as image classification. Despite various optical implementations of electronic neural networks (ENNs), the bulky volume of optical components remains challenging to deploy edge devices, such as Internet of Things peripherals, wearable devices, and camera. To address this problem, we propose a compact lensless optoelectronic convolutional neural network (LOE-CNN) architecture with a lensless optical analog processor utilizing a single optimized diffractive phase mask (DPM) to perform convolution operations without Fourier lens. Comparing the processor with a commercially available NVIDIA A100 Tensor Core GPU in terms of speed and power, indicates the optical computing platform enables to replace the electronic processor in latency reduction and energy savings. Furthermore, we compare the LOE-CNN with two all-electronic neural networks (i.e., fully connected neural network [FC-NN] and convolutional neural network [CNN]) over the Modified National Institute of Standards and Technology (MNIST) dataset and Fashion-MNIST dataset, respectively, and demonstrate that the LOE-CNN can be functionally comparable to existing electronic counterparts in classification performance. My study not only opens up new application prospects for free-space ONNs based on compact lensless single-chip convolution processor, but also facilitates the development of ONNs-based smart devices.
Recovering information of an object hidden behind turbid media has a vast range of applications. People have been trying many ways to achieve this goal. In this paper, we use lensless Fourier transform digital holography and statistical average to retrieve object information from speckle field. The relationship between parameters of ground glass and peak signal-tonoise ratio of reconstructed image is explored by establishing a rotating ground glass interferometric imaging system. The light beam emitted from the laser is divided into two beams by the beam splitter. One beam passes through the rotating ground glass after being reflected by the object (i.e. object light), and the other beam passes through the reflector and convex lens as a reference light, and then interferes with the object light. Finally the interference speckle pattern is captured by a CCD camera. Experiment results show that there is an optimal solution between the rotation speed of ground glass and the peak signal-to-noise ratio of the reconstructed image.This method has the advantages of compact system construction , easy implementation and fast reconstruction, since it does not require phase correction, complex image processing, scanning object or wavefront shaping.
The image distortions caused by the inherent mode dispersion and coupling of the multimode fiber (MMF) lead its output light field to be scattered and prevent it from applicating in endoscopy. Although various wavefront shaping methods have been proposed to overcome these image distortions and form the focused spots through the MMF, they a re usually time-consuming due to the multiple iterations and tedious calculation. In this paper, we present a binary amplitude-only modulation parallel coordinate algorithm for focusing and scanning light through a multimode fiber (MMF) based on the digital micro-mirror device (DMD) in a reference-free multimode fiber imaging system. In principle, our algorithm is capable of efficiently calculating the masks to be added to DMD for yielding a series of tightly focused spots; and for the same number of modulation sub-regions, our method is more than M (the number of focused spots) times faster than the amplitude iterative optimization algorithm. In the experiment, efficient light focusing and scanning at the distal end of the MMF without the iteration process are demonstrated. Furthermore, we demonstrate that the proposed method can also be extended to focus and scan light at multiple planes along the axial direction by just modifying the input wavefront accordingly. We predict the high-speed focusing method through the MMF might have the potential application for fast spot-scanning imaging.
At present, there are some problems in the digital scanning imaging methods of multimode optical fiber, such as poor quality of focused spot, large amount of calculation and long time in the forming process of focused spot. In order to solve the above problems, a parallel phase compensation method based on liquid crystal spatial light modulator (LCSLM) is proposed to achieve fast point-focused scanning of multimode fiber (MMF). The parallel algorithm, which includes the two processes of collecting online specklegrams and calculating offline phase masks, calculates the compensation phase masks so that the phase masks can be loaded on SLM in turn to generate a series of focused spots at different predefined positions, which greatly saves time. Experiments demonstrate that this method doesn’t need many iterations, and the phase compensation mask used for focusing spot at a predefined position on the MMF output facet can be obtained by using a two-step phase-shifting technique, and a series of phase compensation masks can be quickly obtained by using the parallel method, so as to form focused spots at different predefined positions. Experimentally, we obtained 100 focused spots, the average focused efficiency was 30.12%, the average focused diameter was 2.3382μμm, and the quality of the focused spot was improved compared with previous reports.
Light wave becomes extremely distorted when it passes through a turbid medium. Indeed, the inhomogeneity of scattering medium and the mode dispersion of multimode optical fiber (MMF) always distort the propagation of light waves since they divert the propagation direction and disorder the spatial relationship of rays from the object. This becomes a big challenge for the applications of biological tissues endoscopic imaging. To overcome this problem, many methods based on computational optical imaging schemes have been reported and such a research has become a hot topic in recent years. These methods include the computational ghost imaging, the digital phase conjugation, the speckle correlation, the wavefront shaping, and the optical transmission matrix, etc. In this paper, we report our recent works on computational optical imaging based on digital wavefront modulation, which might be useful for the applications of endoscopy. On one hand, we propose a round-trip imaging method based on the optical transmission matrix of scattering medium, where the light wave is distorted twice. The object is recovered directly from the distorted output wave, while no scanning is required during the imaging process; one the other hand, by modulating the amplitude instead of the phase of the incident light wavefront, we propose a high-speed binary amplitude-only modulation method to focus and scan light through an MMF based on the digital micro-mirror device (DMD). This method can also be extended to focus and scan light at multiple planes along the axial direction by just modifying the input wavefront accordingly.
Temperature is an important factor affecting the performance of TO package LD. In order to ensure the safe and stable operation of LD, a temperature control circuit for LD based on PID technology is designed. The MAX1978 and an external PID circuit are used to form a control circuit that drives the thermoelectric cooler (TEC) to achieve control of temperature and the external load can be changed. The system circuit has low power consumption, high integration and high precision,and the circuit can achieve precise control of the LD temperature. Experiment results show that the circuit can achieve effective and stable control of the laser temperature.
The inhomogeneity of scattering medium distorts the propagation of the waves, which has been detrimental to the performance of optical imaging. The operating time of the traditional solutions will be very long as the scanning is necessary during the imaging. A recovery solution based on spatial optical transmission matrix has been proposed. With the acquiring of the spatial optical transmission matrix, the incident object wave will be recovered directly from the distorted transmitted wave, in this way, only a single shot is needed during the imaging. The effectiveness of this method has been proved by the simulation and experiment, the principle is simpler and the algorithm is more efficient, which are beneficial to the imaging through the scattering medium.
Thermal therapy (or hyperthermia) is one of the effective operations for tumor treating and curing. As tumor tissues are more susceptible to heat than normal tissues, in thermal therapy operations, temperature on operation area is a crucial parameter for optimal treating. When the temperature is too low, the tumor tissues cannot be killed; otherwise, the temperature is too high, the operation may damage normal tissues around the tumor. During thermal therapy operation, the heating power is normally supplied by high-frequency EM field, so traditional temperature sensors, such as thermal couples, thermistors, cannot work stably due to EM interference. We present a multi-function endoscope optical fiber temperature sensor system. With this sensor setup based on principle of fluorescence life time, the temperature on operation point is detected in real time. Furthermore, a build-in endoscope centers in the fiber sensor, thus the operation area can be viewed or imaged directly during the operation. This design can navigate the operation, particularly for in vivo operations. The temperature range of the sensor system is 30°C-150°C, the accuracy can achieve to 0.2°C. The imaging fiber buddle is constituted of more than 50k fibers. As the sensor probe is very thin (around 4 mm in diameter), it can also be assembled inside the radiofrequency operation knife. With the presented sensor system in clinic operation physicians can check the temperature in the operation point and view the operation area at the same time.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.