Session Index

The AR device with a holographic curved waveguide combiner is proposed. The device utilized the conjugated recording method to reduce the aberration caused by the curved reflection surfaces. Furthermore, the ghost noise can be avoided by arranging the dimension and position of the HOEs strategically.

 

We proposed a lensfree face recognition system with scattering encryption by means of an optical diffuser. The facial images were encrypted to be unrecognizable amid the imaging chain of a lensfree system. The encrypted intermediate images can be identified by the ResNet18, a neural network model. We experimentally implemented the system and achieve 100% accuracy subject to10 classes(subjects).

 

In this research, we analysis the measurement uncertainty of digital-grating moiré technique for the measurement of lens focal length. The results reveal that the technique can be applied for regular commercial spherical lenses with a measurement uncertainty less than 1.6 mm within a measurement distance of 2 m. The proposed methodology provides a valuable reference for related measurement techniques.

 

This research aims to integrate the two technologies of digital holography and fringe projection profilometry within the same system. The integrated system is capable of measuring the three-dimensional surface of the object. Utilizing the integrated system results in more complete three-dimensional surface information.

 

In this research, we propose an optical configuration to generate a heterodyne light source using a moving grating with constant speed. The results showed that the frequency of the heterodyne signal can be varied with the speed of grating and exhibited similar results of theoretical prediction.

  Preview abstract

The paper proposes a method for measuring small angle based on phase sensitive
total-internal-reflection heterodyne interferometry. When a half-wave plate, a quarter-wave
plate and an analyzer with proper azimuth angles are arranged in the tested arm of the
interferometer, the phase difference caused by total internal reflection of the interference
signal is greatly enhanced. The phase difference can be precisely measured with the
heterodyne interferometric technique, and the associated angle can be estimated. The
experimental results demonstrate the feasibility of this method.

 

In this study present a novel approach using deep learning-based active wavefront aberration correction for beam rotation holographic tomography to acquire three-dimensional image of native biological cell sample. As a result, the proposed algorithm is capable of accurately correct the background aberration and much faster than its original plain algorithm.

 

A novel 3D detection method was proposed by an incoherent holography based fringe projection profilometer. Compare with the conventional fringe projection profilometer, a high resolution quantitative profile can be realized by calculating without complicated adjustment before each detected procedure. The key point of the quantitative measurement is based on an absolute axis reconstruction by using incoherent holographic recording, which can obtain with the profilometer at the same time.

 

This research is a mosquito trap system, which is mainly divided into three parts. Firstly, 365nm UV light is used for visual induction. Secondly, the secondary element lampshade designed for this experiment is used to optimize the light distribution range, so as to be more suitable for attracting mosquitoes. Finally, the odor and CO2 concentration stimulated by UVA and photocatalyst are used to trap the sense of smell, and the trapping system is optimized by combining the above.

 

We proposed a method to calculating far-field wide-angle intensity distribution of diffractive objects using a modification of the Fraunhofer formula. By comparing with the Huygens-Fresnel formula, the proposed method is able to approximate the far-field wide-angle pattern of sub-micro diffractive objects with about 50% less time and the same accuracy.

 

A computer-generated hologram (CGH) comprising multiple CGHs of sliced images of a 3D object produced a volumetric image used to polymerize photosensitive resin. The method features high efficiency, high resolution, rapid fabrication, and good continuity.

 

This study provided a convenient technique for fabricating LC gratings with crystal/quasicrystal/kaleidoscopic patterns. The 2D patterns were generated with a collimated light to illuminate a high-precision mask with multiple apertures regularly distributed on a ring. The 2D LC gratings were electrically tunable and exhibited polarization-independent first-order diffraction.

  Preview abstract

A multi-beam splitter composed of a diffractive optical element ( DOE ) and a beam expander is proposed in the study. Through the optical simulation by ZEMAX and real optical experiments, it can be demonstrated that a laser beam can be split as four beams and their spot size and directions can be controlled by beam expander, in order to meet the requirement of laser printing system for integrated circuits.


 

This study set up the coaxial scanning fringe projection profilometry. And use the direct amplitude extraction method that is a faster way of image processing. Using curve fitting selection to increase the precision of profile reconstruction by choosing proportion correct images data.

 

The phase errors due to the inevitably noise of unstable ambient light, device flicker, or high surface reflectivity in captured fringes, which are non-ignorable in Fringe projection profilometry (FPP). In this paper, a deep learning-based noise detection and its reduction for 3D shape map is proposed. It enables to detect the noisy regions for inspection utilizing captured fringes by the camera. In order to reduce the measurement errors in noisy regions, the inpainting technique of interpolation is proposed. The experimental results demonstrate the method's efficient performance and efficiency without additional fringe reprojection and devices for 3D shape measurement.

 

We present a multimodal holographic microscopy combined digital holographic microscope and Raman spectroscopy to achieve the label-free cell imaging and diagnosis. The preliminary experimental demonstration shows the simultaneous measurement of neuroblastoma cell imaging and spectrum analysis to reveal the potential label-free biological applications.

 

YOLO (You Only Look Once) neural network uses deep learning for real-time multi-object recognition and detection. This study evaluates YOLO using mask detection for mask wearing mandates. The results show that YOLO is able to recognize people with and without masks in real time, and the accuracy is very high.

 

This research proposes an off-axis digital holographic microscopy (DHM) system, which is used to record digital holograms of living cells and obtain cell growth status from the reconstruction phase. This system records and measures cell changes over a long duration.

 

This study proposed a binocular HMD which consists of a planar acrylic waveguide and four HOEs. The diffractive efficiency of the HOEs was expected via coupled wave theory. The FOV for the system achieve to 17.7° in the horizontal direction and 5.3° in the vertical direction.

 

In this study, we demonstrated the heterodyne frequency with an moving grating at constant
speed. The modulated frequency of heterodyne signal was varied with the grating pitch changed. We
successfully showed that two frequencies of heterodyne signal can be obtained by the grating pitch of
0.833 μm and 1.667 μm, respectively.

  Preview abstract

In this paper, we developed an optical license plate automatic recognition system using the YOLO deep learning framework. During the training process, the license plate data set collected from the network database has a total of 1966 pictures, and then 393 pictures are randomly selected as the test set, the remaining images are used as the training set. then use LabelImg software to label and classify all images, and start training the YOLO neural network. finally, through experimental evaluation, it can be seen that the license plate recognition system proposed in this article, the highest mean Average Precision(mAP) is 90%.

 

We applied a high-resolution digital micromirror device (DMD) together with an imaging system at 405 nm to build a high-resolution hologram printer. Because only a small size of hologram fringes can be displayed on the DMD, the hologram data is separated to many tiles, and every time only a tile is projected to the photoresist for exposure. By stitching all the tiles, a large hologram can be produced.