• Korean Journal of Remote Sensing
• /
• v.27 no.2
• /
• pp.141-150
• /
• 2011

Precise image-to-image registration is required to process multi-sensor data together. The purpose of this paper is to develop an algorithm that register between high-resolution optical and SAR images using linear features. As a pre-processing step, initial alignment was fulfilled using manually selected tie points to remove any dislocations caused by scale difference, rotation, and translation of images. Canny edge operator was applied to both images to extract linear features. These features were used to design a cost function that finds matching points based on their similarity. Outliers having larger geometric differences than general matching points were eliminated. The remaining points were used to construct a new transformation model, which was combined the piecewise linear function with the global affine transformation, and applied to increase the accuracy of geometric correction.

• KIPS Transactions on Software and Data Engineering
• /
• v.5 no.12
• /
• pp.671-676
• /
• 2016

Recently, according to the limelight of Fintech, simple payment using biometric at smartphone is widely used. In this paper, we propose a new contactless biometric method using thumb image without additional sensors unlike previous biometrics such as fingerprint, iris, and vein recognition. In our method, length, width, and skin texture information are used as features. For that, illumination normalization, skin region segmentation, size normalization and alignment procedures are sequentially performed from the captured thumb image. Then, correlation coefficient is calculated for similarity measurement. To analyze recognition accuracy, genuine and imposter matchings are performed. At result, we confirmed the FAR of 1.68% at the FRR of 1.55%. In here, because the distribution of imposter matching is almost normal distribution, our method has the advantage of low FAR. That is, because 0% FAR can be achieved at the FRR of 15%, the proposed method is enough to 1:1 matching for payment verification.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.40 no.11
• /
• pp.997-1003
• /
• 2012

GNSS and ARS are the most common sensors in low-end UAVs. However, these sensors are vulnerable to built-in errors and cannot measure the body heading independently. The GNSS/INS cannot fully compensate the IMU errors in initial alignment process and rectilinear flights. For an unmanned helicopter, a magnetometer can be more useful than any other sensors to obtain heading information. However, the electric motor which drives small helicopter UAV keeps the magnetometer from reading the pure magnetotelluric vector. This paper shows the effects of electric motor on the magnetometer readings, and presents a method to compensate the effects. The results are verified with flight test data. The simulation and experimental results in this paper proves that aiding GNSS/INS with magnetometer increases observability and improves accuracy.

• Journal of Digital Contents Society
• /
• v.18 no.1
• /
• pp.93-99
• /
• 2017

A hybrid approach using Long Short Term Memory (LSTM) Recurrent Neural Network (RNN) has showed great improvement in speech recognition accuracy. For training acoustic model based on hybrid approach, it requires forced alignment of HMM state sequence from Gaussian Mixture Model (GMM)-Hidden Markov Model (HMM). However, high computation time for training GMM-HMM is required. This paper proposes an end-to-end approach for LSTM RNN-based Korean speech recognition to improve learning speed. A Connectionist Temporal Classification (CTC) algorithm is proposed to implement this approach. The proposed method showed almost equal performance in recognition rate, while the learning speed is 1.27 times faster.

• The Journal of Advanced Prosthodontics
• /
• v.10 no.4
• /
• pp.321-327
• /
• 2018

PURPOSE. The purpose of this study was to compare the cutting method and the lamination method to investigate whether the CAD data of the proposed inlay shape are machined correctly. MATERIALS AND METHODS. The Mesial-Occlusal shape of the inlay was modeled by changing the stereolithography (STL). Each group used SLS (metal powder) or SLA (photocurable resin) in the additive method, and wax or zirconia in the subtractive method (n=10 per group, total n=40). Three-dimensional (3D) analysis program (Geomagic Control X inspection software; 3D systems) was used for the alignment and analysis. The root mean square (RMS) in the 2D plane state was measured within $50{\mu}m$ radius of eight comparison measuring points (CMP). Differences were analyzed using one-way analysis of variance and post-hoc Tukey's test were used (${\alpha}=.05$). RESULTS. There was a significant difference in RMS only in SLA and SLS of 2D section (P<.05). In CMP mean, CMP 4 ($-5.3{\pm}46.7{\mu}m$) had a value closest to 0, while CMP 6 ($20.1{\pm}42.4{\mu}m$) and CMP 1 ($-89.2{\pm}61.4{\mu}m$) had the greatest positive value and the greatest negative value, respectively. CONCLUSION. Since the errors obtained from the study do not exceed the clinically acceptable values, the lamination method and the cutting method can be used clinically.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.12 no.3
• /
• pp.440-449
• /
• 1988

Small vibration displacements may be measured by optical interferometers, based on the Michelson method. The standard Michelson interferometer works well when the mirror displacements are relatively large compared to the optical wavelength. But it does not work for displacements less than approximately a quater of optical wavelength. Several multiple reflection laser interferometers, simply modified standard Michelson interferometer, have been developed to decrease the minimum detectable limits. Among these a relatively simple and easy multiple reflection system is used to measure the small vibration displacements. This multiple reflection system is constructed with a right angle prism and a convex lens. Therefore this system makes it possible to measure a vibration displacement of the small area on the vibrating structure. The fringe interpolation method and curve fitting method are used to determine accurately the small vibration displacements from the measured interference fringe patterns. Also computer simulation technique is used to check the accuracies of these method. According to the results of the computer simulation technique, the curve fitting method is more accurate than the fringe interpolation method. The optically measured results are in good agreement with those of the standard accelerometer with high accuracy and it is possible to measure the peak vibration displacement as small as 9.01nm using multiple reflection system and curve fitting method.

• Journal of the Korean Society of Manufacturing Process Engineers
• /
• v.14 no.4
• /
• pp.151-159
• /
• 2015

In this study, we developed an automated contact-type thickness measurement machine that continuously and precisely measures the thickness of a PCB module product using multi-LVDT sensors. The system contains a measurement part to automatically measure the thickness in real time according to the set conditions with an alignment supply unit and unloading unit to separate OK and NG products. The sensors were calibrated before assembly in the measuring machine, and precision and accuracy performance tests were also performed to reduce uncertainty errors in the measurement machine. In the calibration test, the precision errors of the LVDT sensor were determined to be $1-3{\mu}m$ as 0.1% at the measuring range. A measurement error of 0.8 mm and 1.0 mm thickness test standards were found to be $1{\mu}m$ and $4{\mu}m$, and the standard deviations of two 1.0 mm products were measured as $14{\mu}m$ and $8{\mu}m$, respectively. In the measurement system analysis, the accuracies of test PCB standards were found to be $2{\mu}m$ and $3{\mu}m$, respectively. From the results of gage repeatability and reproducibility (R & R) crossed, we found that the machine is suitable for the measurement and process control in the mass production line as 7.92% of total gage R & R and in seven distinct categories. The maximum operating speed was limited at 13 pcs/min, showing a value good enough to measure.

• Proceedings of the Korean Society of Medical Physics Conference
• /
• 2002.09a
• /
• pp.189-191
• /
• 2002

In the proton therapy using a gantry system, periodical verification of iso-center position is very important to assure precision of patient positioning system at any gantry angles in proton treatment. In the gantry system, there are three different types of iso-center; 1) in a geometrical view, 2) in an X-ray beam's eye view, 3) in a proton beam's eye view. Idealistically, they would be an identical point. They could, however, be different points. It may be a source of errors in patient positioning. At PMRC, we have established a system of verification for iso-center positions using a stainless ball of 2-cm in diameter and an imaging plate. This system provides the relation among a center of a patient target position, a center of proton irradiation field, and/or a center of X-ray field in accuracy of 50$\square$m in the 2) and 3) views, as images of a center of the stainless ball and a center of a 100 mm${\times}$100 mm-aperture brass collimator recorded on the imaging plate, which is setup at 1-cm behind the ball. In addition, it provides simultaneously the images of the ball and the collimator on an imaging intensifier (II), which is setup downstream of the proton or X-ray beam. We present a method of quality assurance (QA) for calibration of iso-center position in a rotation gantry system at PMRC and the performance of this system. A proton beam position on the 1$\^$st/ scatterer in the nozzle of the gantry affects less sensitive (reduced by a factor of 1/5) to the results of the iso-center position. The effect is systematically correctable. The effect of the nozzle (or the collimator) position is less than 0.5 mm at the maximum extraction (390 mm).

• Korean Journal of Optics and Photonics
• /
• v.15 no.6
• /
• pp.583-590
• /
• 2004

The autofocus is one of the important processes in the automated vision inspection or measurements using optical microscopes, because it influences the measuring accuracy. In this paper, we used the confocal microscope configuration based on not a pinhole but a single-mode optical fiber. A single mode fiber has the functions of source and detector by applying the reciprocal scheme. As a result, we acquired a simple system configuration and easy alignment of the optical axis. Also, we embodied a fast autofocus system by acquiring the focus error signal through a source modulation technique. The source modulation technique can effectively reduce physical disturbances compared with objective lens modulation, and it is easily applicable to general optical microscopes. The focus error signal was measured with respect to the modulation amplitude, reflectance of the specimen and inclination angle of the measuring surface. The performance of the proposed autofocus system was verified through autofocusing flat mirror surface. In addition, we confirmed that source modulation rarely degrades the depth resolution by the comparison between the FWHMs of axial response curves.

• Journal of Korean Society for Geospatial Information Science
• /
• v.21 no.2
• /
• pp.35-43
• /
• 2013

Georeferenced SLI(Street-Level Imagery) services such as Google Streetview, which contain abundant information about the real world, can increase its applicability substantially through conflation with other spatial datasets. For this purpose, we propose a method to improve a correspondence of building region to combine building information more accurately. First, the spatial inconsistency between SLI and vector map is removed by alignment based on road intersections. Then, visible building regions are searched from the spatial inconsistency-removed vector map, and the optimal corresponding building areas are determined in the SLI scene using the visible regions as seed information. The experimental results demonstrated that our method had improved the accuracy of building region correspondence by about 8%. Therefore, our method can be utilized effectively for enhancement of conflation service based on the SLI.