Improving the reliability of deep learning computational ghost imaging with prediction uncertainty based on neighborhood feature maps.

Defect inspection is required in various fields, and many researchers have attempted deep-learning algorithms for inspections. Deep-learning algorithms have advantages in terms of accuracy and measurement time; however, the reliability of deep-learning outputs is problematic in precision measurements. This study demonstrates that iterative estimation using neighboring feature maps can evaluate the uncertainty of the outputs and shows that unconfident error predictions have higher uncertainties. In ghost imaging using deep learning, the experimental results show that removing outputs with higher uncertainties improves the accuracy by approximately 15.7%.

Relationship between the kernel size of a convolutional layer and the optical point spread function in ghost imaging using deep learning for identifying defect locations.

We explore the contribution of convolutional neural networks to correcting for the effect of the point spread function (PSF) of the optics when applying ghost imaging (GI) combined with deep learning to identify defect positions in materials. GI can be accelerated by combining GI and deep learning. However, no method has been established for determining the relevant model parameters. A simple model with different kernel sizes was built. Its accuracy was evaluated for data containing the effects of different PSFs. Numerical analysis and empirical experiments demonstrate that the accuracy of defect identification improved by matching the kernel size with the PSF of the optics.

Noise-robust deep learning ghost imaging using a non-overlapping pattern for defect position mapping.

Defect detection requires highly sensitive and robust inspection methods. This study shows that non-overlapping illumination patterns can improve the noise robustness of deep learning ghost imaging (DLGI) without modifying the convolutional neural network (CNN). Ghost imaging (GI) can be accelerated by combining GI and deep learning. However, the robustness of DLGI decreases in exchange for higher speed. Using non-overlapping patterns can decrease the noise effects in the input data to the CNN. This study evaluates the DLGI robustness by using non-overlapping patterns generated based on binary notation. The results show that non-overlapping patterns improve the position accuracy by up to 51%, enabling the detection of defect positions with higher accuracy in noisy environments.

Fabrication of three-dimensional high-aspect-ratio structures by oblique-incidence Talbot lithography.

Developing a suitable production method for three-dimensional periodic nanostructures with high aspect ratios is a subject of growing interest. For mass production, Talbot lithography offers many advantages. However, one disadvantage is that the minimum period of the light intensity distribution is limited by the period of the diffraction grating used. To enhance the aspect ratio of fabricated nanostructures, in the present study we focus on multiple wave interference between diffracted waves created using the Talbot effect. We propose a unique exposure method to generate multiple wave interference between adjacent diffraction orders by controlling the angle of incidence of an ultraviolet (UV) light source. Using finite-difference time-domain simulations, we obtain fringe patterns with a sub-wavelength period using a one-dimensional periodic grating mask. Moreover, we demonstrate the practical application of this approach by using UV lithography to fabricate sub-wavelength periodic photopolymer-based structures with an aspect ratio of 30 in millimeter-scale areas, indicating its suitability for mass production.

Microdisplacement sensor using an optically trapped microprobe based on the interference scale.

Positioning technology is one of the most important technologies for developing microsystems. In particular, displacement sensors are necessary for positioning devices with nanoscale accuracy. In this study, we propose a new displacement sensor that uses an interference scale as a linear scale and a laser-trapped microsphere as a sensing probe. This sensor has a wide measuring range, high resolution, and accessibility for narrow target areas. A glass microsphere was optically trapped by means of the laser trapping technique. Between the target surface and the probe, an interference scale was generated along the optical axis. The scale origin was fixed on the target surface. The distance between the probe and the target surface could be measured in terms of the shift in the interference scale. This study investigated the fundamental performance of the sensor. The resolution and accuracy of the sensor were 10 and +/-50 nm, respectively; these values could be improved by using trapping lasers having shorter wavelengths. The measurable range was 250 microm. This sensor can provide useful displacement information from a target area having dimensions smaller than 15 microm. In addition, the displacement sensor can measure the distance even for surfaces inclined at angles less than 15 degrees; thus, a flexible arrangement can be used to carry out measurements. In addition, the direction of displacement can be identified.

Measurement of axial and transverse trapping stiffness of optical tweezers in air using a radially polarized beam.

The trapping efficiency and stiffness of optical tweezers using radial polarization are evaluated; the ray-tracing method and a proposed measurement method are used for numerical and experimental analyses, respectively. The maximum axial trapping efficiency with radial polarization is 1.84 times that with linear polarization, while the maximum transverse trapping efficiency decreases by 0.58 times. Further, the axial and transverse trapping efficiencies are found to be 1.19 times larger and 0.83 times smaller, respectively, than the values with linear polarization. From the experiments, the axial and transverse stiffness values are 1.2 times larger and 0.8 times smaller, respectively, with radial polarization. Hence, radial polarization enhances the axial trapping properties while reducing the transverse trapping properties.

Probing technique using circular motion of a microsphere controlled by optical pressure for a nanocoordinate measuring machine.

A new surface probing technique using the circular motion of an optically-trapped microsphere is proposed for a nanocoordinate measuring system. The probe sphere is oscillated circularly in the plane perpendicular to the probe axis and the circular orbit of the probe sphere is monitored for the detection of the position and normal vector direction of the surface. The principle of detection is based on changes in the circular orbit of the microsphere. When the probe approaches a work surface, the orbit of the probe sphere becomes elliptical. The minor-axis length and the minor-axis angle of the ellipse are then used as parameters to detect the position and normal vector direction of the surface, respectively. In this study, the circular motion probe is shown to have a resolution of position detection of 39 nm, and the accuracy of measuring a normal vector to the surface is on the order of 3 degrees.