American Sign Language Recognition and Translation Using Perception Neuron Wearable Inertial Motion Capture System.

Sign language is designed as a natural communication method to convey messages among the deaf community. In the study of sign language recognition through wearable sensors, the data sources are limited, and the data acquisition process is complex. This research aims to collect an American sign language dataset with a wearable inertial motion capture system and realize the recognition and end-to-end translation of sign language sentences with deep learning models. In this work, a dataset consisting of 300 commonly used sentences is gathered from 3 volunteers. In the design of the recognition network, the model mainly consists of three layers: convolutional neural network, bi-directional long short-term memory, and connectionist temporal classification. The model achieves accuracy rates of 99.07% in word-level evaluation and 97.34% in sentence-level evaluation. In the design of the translation network, the encoder-decoder structured model is mainly based on long short-term memory with global attention. The word error rate of end-to-end translation is 16.63%. The proposed method has the potential to recognize more sign language sentences with reliable inertial data from the device.

Agar-based optical sensors for electric current measurements.

Biodegradable optical waveguides are breakthrough technologies to light delivery and sensing in biomedical and environmental applications. Agar emerges as an edible, soft, low-cost, and renewable alternative to traditional biopolymers, presenting remarkable optical and mechanical characteristics. Previous works introduced agar-made optical fibers for chemical measurements based on their inherent response to humidity and surrounding concentration. Therefore, we propose, for the first time, an all-optical, biodegradable electric current sensor. As flowing charges heat the agar matrix and modulate its refractive index, we connect the optical device to a DC voltage source using pin headers and excite the agar sample with coherent light to project spatiotemporally deviating speckle fields. Experiments proceeded with spheres and no-core fibers comprising 2 wt% agar/water. Once the increasing current stimulates the speckles' motion, we acquire such images with a camera and evaluate their correlation coefficients, yielding exponential decay-like functions whose time constants provide the input amperage. Furthermore, the light granules follow the polarization of the applied voltage drop, providing visual information about the current direction. The results indicate a maximum resolution of [Formula: see text]0.4 [Formula: see text]A for electrical stimuli [Formula: see text] 100 [Formula: see text]A, which fulfills the requirements for bioelectrical signal assessment.

1000-volume/s high-speed volumetric display for high-speed HMD.

In this paper, we propose a high-speed volumetric display principle that can solve two problems faced by three-dimensional displays using the parallax stereo principle (namely, the vergence-accommodation conflict and display latency) and we report evaluation results. The proposed display method can update a set of images at different depths at 1000 Hz and is consistent with accommodation. The method selects the depth position in microseconds by combining a high-speed variable-focus lens that vibrates at about 69 kHz and sub-microsecond control of illumination light using an LED. By turning on the LED for only a few hundred nanoseconds when the refractive power of the lens is at a certain value, an image can be presented with this specific refractive power. The optical system is combined with a DMD to form an image at each depth. 3D information consisting of multiple planes in the depth direction can be presented at a high refresh rate by switching the images and changing the refractive power at high speed. A proof-of-concept system was developed to show the validity of the proposed display principle. The system successfully displayed 3D information consisting of six binary images at an update rate of 1000 volume/s.

Quasi-simultaneous multi-focus imaging using a lock-in pixel image sensor and TAG lens.

In this paper, a quasi-simultaneous multi-focus imaging technique named simulfocus imaging is reported. This technique was developed for measuring an entire object distributed in the depth direction beyond the depth of field (DOF) with high resolution in a single shot. Simulfocus imaging can acquire multiple focal planes in one shot by synchronizing a tunable acoustic gradient index (TAG) lens and a lock-in pixel image sensor. The TAG lens is a tunable-focus lens whose focal position can be changed at a high speed of several tens to several hundreds of kilohertz. The lock-in pixel image sensor is a special image sensor that can execute multiple exposures at an arbitrary timing during a single shooting. The sensor includes a number of photoelectron storage units in each pixel, and the units where the photoelectrons generated by each exposure are stored can be freely selected. Since an image can be acquired for a single storage unit, and the lock-in pixel image sensor has a number of storage units, the lock-in pixel image sensor can acquire multiple images in one shot. By assigning a specific exposure timing to each unit and synchronizing the exposure timing with the focus fluctuation of the TAG lens, it is possible to simultaneously acquire images in different focal planes. To evaluate the system, we conducted experiments to show the effectiveness of simulfocus imaging in microscope and telescope configurations. From the experimental results, it was confirmed that simulfocus was effective in both configurations.

Agarose-based structured optical fibre.

Biocompatible and resorbable optical fibres emerge as promising technologies for in vivo applications like imaging, light delivery for phototherapy and optogenetics, and localised drug-delivery, as well as for biochemical sensing, wherein the probe can be implanted and then completely absorbed by the organism. Biodegradable waveguides based on glasses, hydrogels, and silk have been reported, but most of these devices rely on complex fabrication procedures. In this sense, this paper proposes a novel structured optical fibre made of agarose, a transparent, edible material used in culture media and tissue engineering. The fibre is obtained by pouring food-grade agar into a mould with stacked rods, forming a solid core surrounded by air holes in which the refractive index and fibre geometry can be tailored by choosing the agarose solution composition and mould design, respectively. Besides exhibiting practical transmittance at 633 nm in relation to other hydrogel waveguides, the fibre is also validated for chemical sensing either by detecting volume changes due to agar swelling/dehydration or modulating the transmitted light by inserting fluids into the air holes. Therefore, the proposed agarose-based structured optical fibre is an easy-to-fabricate, versatile technology with possible applications for medical imaging and in vivo biochemical sensing.

Fast Volumetric Feedback under Microscope by Temporally Coded Exposure Camera.

We developed a temporally coded exposure (TeCE) camera that can cope with high-speed focus variations of a tunable acoustic gradient index (TAG) lens. The TeCE camera can execute a very short exposure multiple times at an arbitrary timing during one shot. Furthermore, by accumulating the photoelectrons generated by each exposure, it is possible to maintain the brightness even with a short exposure time. By synchronously driving the TeCE camera and the TAG lens, different focal planes of an observation target can be acquired at high speed. As a result, high-speed three-dimensional measurement becomes possible, and this can be used for feedback of three-dimensional information. In the work described in this paper, we conducted a focus tracking experiment to evaluate the feedback performance of the TeCE camera. From the experimental results, we confirmed the feedback capability of the TeCE camera.

High-speed microscopy with an electrically tunable lens to image the dynamics of in vivo molecular complexes.

We provide an evaluation for an electrically tunable lens (ETL), combined with a microscope system, from the viewpoint of tracking intracellular protein complexes. We measured the correlation between the quantitative axial focus shift and the control current for ETL, and determined the stabilization time for refocusing to evaluate the electrical focusing behaviour of our system. We also confirmed that the change of relative magnification by the lens and associated resolution does not influence the ability to find intracellular targets. By applying the ETL system to observe intracellular structures and protein complexes, we confirmed that this system can obtain 10 nm order z-stacks, within video rate, while maintaining the quality of images and that this system has sufficient optical performance to detect the molecules.

An improved low-optical-power variable focus lens with a large aperture.

We report an improved method of fabricating a variable focus lens in which an in-plane pretension force is applied to a membrane. This method realized a lens with a large optical aperture and high performance in a low-optical-power region. The method was verified by comparing membranes in a simulation using the finite element method. A prototype with a 26 mm-diameter aperture was fabricated, and the wavefront behavior was measured by using a Shack-Hartmann sensor. Thanks to the in-plane pretension force, the lens achieved an infinite focal length with a wavefront error of 105.1 nm root mean square.

How to track spermatozoa using high-speed visual feedback.

In this paper, we report how to track quickly and vigorously swimming ascidian spermatozoa using high-speed visual feedback at a frame rate of 1 kHz. Ascidian spermatozoa swim as fast as 300 microm/s by rotating their flagella 50 times/s. This vigorous swimming style has prevented stable image observation and made it difficult to track them reliably with our previously developed visual tracking system. Here, we describe how we overcame these problems using image processing techniques to achieve stable tracking of fast, small ascidian spermatozoa for more than 180 s using high-speed visual feedback.

A physical model for galvanotaxis of Paramecium cell.

We propose a qualitative physical model of galvanotaxis of Paramecium cells using a bottom-up approach to link the microscopic ciliary motion and the macroscopic behavior of the cells. From the characteristic pattern of ciliary motion called the Ludloff phenomenon, the torque that orients the cell toward the cathode is derived mathematically. Dynamical equations of motion are derived and their stability is discussed. In numerical simulations using our model, cells exhibit realistic behavior, such as U-turns, like real cells.

Variable-focus lens with 1-kHz bandwidth.

This paper proposes a variable-focus lens with 1-kHz bandwidth. The lens transforms its shape rapidly using the liquid pressure generated by a piezo stack actuator. This mechanism also includes a built-in motion amplifier with high bandwidth to compensate for the short working range of the piezo stack actuator. Prototypes have been developed to validate the proposed design. A 1-kHz bandwidth of the lenses was confirmed by measuring the frequency responses. Refractive power ranging from -1/167 to 1/129 mm(-1) and a maximum resolution of 12.3 cycles/mm were attained.