Session Index

Scratches at specific locations on a smooth surface can produce three-dimensional bright spots, which is called scratch stereograms. Here, we took points from the input 3D model and calculated the location and orientation of the scratches. The scratch stereogram is fast fabricated on a plate by CNC engraving.

  Preview abstract

This paper presents a prototype of near-eye display device, containing a fiber source, a liquid-crystal-on-silicon spatial light modulator (LCoS-SLM), and a holographic optical lens (HOL). The total length of the presented apparatus from a fiber source to the LCoS-SLM is about 12 cm and the projected image was observed 25 cm behind the LCoS-SLM. The image is created from a diffractive optical element (DOE) displayed on the LCoS-SLM. The DOE is calculated by iterative Fourier transform algorithm (IFTA) followed by multiplied with a quadratic phase term to mimic a convergent lens.

  Preview abstract

This study provides a method for tracking features on cardiac surface vessels for determining the heart rate by applying convolutional neural networks (CNN). We designed a complete procedure to accomplish this task, which includes image preprocessing, CNN model training, and sliding windows algorithm for feature finding. We tracked two features accurately and estimated the heart rate by analyzing the distance between them. This research explores the implementation of deep learning for extracting heart behaviors from a series of recorded medical images and the potential advancements in cardiac dynamic analysis and surgery monitoring.

  Preview abstract

In this work, we perform a systematic study of different deconvolution techniques, namely Wiener, Richardson-Lucy (RL), and the innovative Richardson-Lucy with total variation (RLTV), and compare their performance. Our results indicate that RL deconvolution outperforms Wiener regarding contrast enhancement and drastically improves image quality. RLTV is effective, efficient, and accurate, which may find important applications in high-resolution imaging.

  Preview abstract

The Moiré effect is a frequently employed image processing technique in optical measurements. In this paper, we employ MATLAB programming to numerically simulate and analyze the interference patterns of Moiré fringes based on the relative rotational angles and relative period ratios. Gaining insight into these Moiré pattern characteristics will aid in the application and development of digital Moiré effect measurement technologies.

  Preview abstract

In this study, a 532nm green laser projects diffractive optical elements (DOE) in a spatial light modulator (SLM) to produce multiple 2D images separated in different spatial layers. Each layer's DOE was created using the iterative angular spectrum algorithm (IASA) for imaging in optical near field. Several techniques were employed to improve the image quality: Image shifting separates target shapes from SLM's reflection, temporal superposition reduces the speckle noise, and grayscale integration improves object continuity. The process shapes target objects using high-transparency resin, offering benefits of internal visualization, vertical continuity, and efficiency despite limited vertical variation and operational experience requirements.

  Preview abstract

Defects in compound semiconductor wafers may affect the lifetime as well as decrease the performance of the power devices. Therefore, in this study, we demonstrated a computational refocusing algorithm in combination with a high-resolution swept-source optical coherence tomography (SS-OCT) system setup to enhance the contrast of the defect images in silicon carbide (SiC) wafer. The location, distribution, and size of defects, including macro pits, stacking faults, micropipes, and particle inclusions, were analyzed quantitatively.

  Preview abstract

This work focuses on using deep neural networks for reconstructing structural parameters from spectral information. RCWA MATLAB code is used to simulate spectral signals (Psi and Delta) for different FinFET-Fin parameters. Simulated ellipsometric scatterometer data is utilized to train the inverse neural network (INN) by using PyTorch. The INN extracts critical features through convolution, aiming to predict accurate structural parameters by comparing outputs with actual values. On the other hand, contrary to INN, the forward neural network (FNN) predicts spectra based on structural parameters through transposed convolutions.

  Preview abstract

The edge contrast enhancement can be explained using Hilbert and fractional Hilbert transforms. Implementation of spiral phase filtering can be performed experimentally or by image post-processing. Here, we performed a systematic study to demonstrate the effect of the order of phase singularity of the spiral phase mask on edge contrast enhancement. The effect is demonstrated using the simulation. It is observed that by continuously varying the order of phase singularity, the edge enhancement can be altered in a systematic manner. A continuous change of enhancement can be achieved from an anisotropic left to perfectly isotropic to an anisotropic right edge.

  Preview abstract

Using a 532nm green light laser as the recording light for intermittent exposure of the PQ/PMMA material through the two-beam interference method. This process can create periodic refractive index change structures in PQ/PMMA. Subsequently, detecting the diffraction efficiency by changing the light intensity, exposure time, and diffusion time, while also determining how they affect the diffraction efficiency. Finally, with the above experiment, the diffusion coefficient of PQ molecules will be found.

  Preview abstract

In this work, we demonstrate a technique to detect the positions of vortices using a digital micromirror device as a scanning triangular aperture and by measuring the intensity at the center of a pattern using a digital pinhole.

  Preview abstract