• Archives of Craniofacial Surgery
• /
• v.22 no.2
• /
• pp.78-84
• /
• 2021

Background: Facial feminization surgery and malarplasty require information concerning facial features in the malar area. Such information varies as a function of sex and race. The objectives of this study aimed to quantitatively evaluate the location of malar prominence across sexes in the Southeast Asian population, and identify sex-specific differences in malar prominence using a combination of two-dimensional (2D) computed tomography (CT) and three-dimensional (3D) CT. Methods: The location of malar prominence was evaluated in 101 Thai adults, consisting of 52 men and 49 women. This study used both 2D CT and 3D CT to achieve greater accuracy, in which 2D CT was used to measure malar distance, malar summit width, facial width, and malar summitto-facial width ratio whereas 3D CT was used to evaluate the positional relationship between the zygomatic summit and four reference points of the zygoma. Results: The malar summit was positioned more laterally in males (p< 0.01) and was more projected in females (p= 0.01). The other 2D-parameters were wider in males. The ratio between the malar summit width and facial width showed similar results for both sexes. The vertical dimension did not show any statistically significant differences; however, a higher summit position was observed in males. Conclusion: The zygomatic summit is positioned more laterally in males and is more projected in females. However, the ratio was similar, which indicates that the male cranium is larger in size. Based on the results in this study, when facial feminization surgery or malarplasty is performed on a Southeast Asian patient, the malar bone should be reduced horizontally and moved forward for better outcomes.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.20 no.1
• /
• pp.477-483
• /
• 2019

In this paper, we propose vanishing-point position-improvement algorithms by using multiple images and a homography matrix. Vanishing points can be detected from a single image, but the positions of detected vanishing points can be improved if we adjust their positions by using information from multiple images. More accurate indoor space information detection is possible through vanishing points with improved positional accuracy. To adjust a position, we take three images and detect the information, detect the homography matrix between the walls of the images, and convert the vanishing point positions using the detected homography. Finally, we find an optimal position among the converted vanishing points and improve the vanishing point position. The experimental results compared an existing algorithm and the proposed algorithm. With the proposed algorithm, we confirmed that the error angle to the vanishing point position was reduced by about 1.62%, and more accurate vanishing point detection was possible. In addition, we can confirm that the layout detected by using improved vanishing points through the proposed algorithm is more accurate than the result from the existing algorithm.

• Journal of Radiation Protection and Research
• /
• v.44 no.1
• /
• pp.32-42
• /
• 2019

Background: There have been much efforts to develop the proper and realistic machine Quality Assurance (QA) reflecting on real Volumetric Modulated Arc Therapy (VMAT) plan. In this work we propose and test a special VMAT plan of plan-class specific (pcsr) QA, as a machine QA so that it might be a good solution to supplement weak point of present machine QA to make it more realistic for VMAT treatment. Materials and Methods: We divided human body into 5 treatment sites: brain, head and neck, chest, abdomen, and pelvis. One plan for each treatment site was selected from real VMAT cases and contours were mapped into the computational human phantom where the same plan as real VMAT plan was created and called plan-class specific reference (pcsr) QA plan. We delivered this pcsr QA plan on a daily basis over the full research period and tracked how much MLC movement and dosimetric error occurred in regular delivery. Several real patients under treatments were also tracked to test the usefulness of pcsr QA through comparisons between them. We used dynalog file viewer (DFV) and Dynalog file to analyze position and speed of individual MLC leaf. The gamma pass rate from portal dosimetry for different gamma criteria was analyzed to evaluate analyze dosimetric accuracy. Results and Discussion: The maxRMS of MLC position error for all plans were all within the tolerance limit of < 0.35 cm and the positional variation of maxPEs for both pcsr and real plans were observed very stable over the research session. Daily variations of maxRMS of MLC speed error and gamma pass rate for real VMAT plans were observed very comparable to those in their pcsr plans in good acceptable fluctuation. Conclusion: We believe that the newly proposed pcsr QA would be useful and helpful to predict the mid-term quality of real VMAT treatment delivery.

• Journal of Korean Society of Forest Science
• /
• v.110 no.1
• /
• pp.72-81
• /
• 2021

We developed smartphone application and web application server to acquire and effectively manage tree measurement information. Smartphone applications can measure tree height, diameter at breast height (DBH), azimuth, altitude, slope, and positional coordinates using augmented reality (Google AR core) and motion sensors. The web application server effectively manages and stores measurement information. To evaluate the accuracy of information acquired using a smartphone, 90 Korean pine trees (Pinus koraiensis) were randomly selected from a natural mixed forest, with a total of 90 representative trees randomly collected from a natural mixed forest. Then, height and DBH were measured using a Haglof Vertex Laser Hypsometer and caliper. Comparisons of the results indicated significant results at the 95% level and a very high average correlation of 0.972 for both tree height and DBH. In terms of DBH, the average errors were 0.6745 cm and 1.0139 cm for artificial coniferous and natural mixed forests, respectively.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.05a
• /
• pp.413-415
• /
• 2022

Since human joints can be used as useful information for analyzing human behavior as a component of the human body, many studies have been conducted on human action recognition using joint information. However, it is a very complex problem to recognize human action that changes every moment using only each independent joint information. Therefore, an additional information extraction method to be used for learning and an algorithm that considers the current state based on the past state are needed. In this paper, we propose a human action recognition technique considering the positional relationship of connected joints and the change of the position of each joint over time. Using the pre-trained joint extraction model, position information of each joint is obtained, and bone information is extracted using the difference vector between the connected joints. In addition, a simplified neural network is constructed according to the two types of inputs, and spatio-temporal features are extracted by adding LSTM. As a result of the experiment using a dataset consisting of 9 behaviors, it was confirmed that when the action recognition accuracy was measured considering the temporal and spatial relationship features of each joint, it showed superior performance compared to the result using only single joint information.

• Korean Journal of Remote Sensing
• /
• v.39 no.6_3
• /
• pp.1707-1720
• /
• 2023

With the increase in satellite time series data, the utility of remote sensing data is growing. In the analysis of time series data, the relative positional accuracy between images has a significant impact on the results, making image registration essential for correction. In recent years, research on image registration has been increasing by applying deep learning, which outperforms existing image registration algorithms. To train deep learning-based registration models, a large number of image pairs are required. Additionally, creating a correlation map between the data of existing deep learning models and applying additional computations to extract registration points is inefficient. To overcome these drawbacks, this study developed a data augmentation technique for training image registration models and applied it to OffsetNet, a registration model that predicts the offset amount itself, to perform image registration for KOMSAT-2, -3, and -3A. The results of the model training showed that OffsetNet accurately predicted the offset amount for the test data, enabling effective registration of the master and slave images.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.31 no.3
• /
• pp.193-198
• /
• 2013

This study proposed a map registration method of a building construction plan drawing with shape matching of cadastral parcel polygon. In general, the drawing contains information about a building boundary and a cadastral parcel boundary. The shape of this cadastral parcel boundary should be same as that of the corresponding parcel polygon object in the KLIS continuous cadastral map. Thus, shape matching between two parcel boundary polygons from the drawing and cadastral map could present transformation parameters. Translation and scaling amounts could be obtained by difference of centroid coordinates and area ratio of the polygons, respectively. Rotation amount could be obtained by the rotation that presents the minimum Turning function dissimilarity of the polygons. The proposed method was applied for building construction plan drawings in eAIS for an urban area in Suwon. To assess positional accuracy of map registration, building polygons in registered drawings and aerial photos were compared. According to the accuracy of the cadastral map which is the reference dataset of the proposed method, the RMSE of corresponding buildings' corners was 0.95m and 2.37m in new and old urban areas, respectively.

• Journal of the Korean Society of Radiology
• /
• v.13 no.7
• /
• pp.901-907
• /
• 2019

The aim of this study is to evaluate the ability of target localizations of Leksell GammaPlan(LGP) in Gamma Knife Subthalamotomy(or Pallidotomy, Thalamotomy) of functional diseases. To evaluate the accuracy of LGP's location settings, the difference Δr of the target coordinates calculated by LGP (or LSP) and author's algorithm was reviewed for 10 patients who underwent Deep Brain Stimulation(DBS) surgery. Δr ranged from 0.0244663 mm to 0.107961 mm. The average of Δr was 0.054398 mm. Transformation matrix between stereotactic space and brain atlas space was calculated using PseudoInverse or Singular Value Decomposition of Mathematica to determine the positional relationship between two coordinate systems. Despite the precise frame positioning, the misalignment of yaw from -3.44739 degree to 1.82243 degree, pitch from -4.57212 degree to 0.692063 degree, and rolls from -6.38239 degree to 7.21426 degree appeared. In conclusion, a simple in-house algorithm was used to test the accuracy for location settings of LGP(or LSP) in Gamma Knife platform and the possibility for Gamma Knife Subthalamotomy. The functional diseases can be treated with Gamma Knife Radiosurgery with safety and efficacy. In the future, the proposed algorithm for target localizations' QA will be a great contributor to movement disorders' treatment of several Gamma Knife Centers.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.20 no.9
• /
• pp.465-473
• /
• 2019

This paper is meaningful as the first case where a 4 - sided hull-fixed phased array radar was installed on a mast separated from Korea and the alignment was verified. The mechanical alignment method was studied for accurately mounting two separate masts for naval ships and the 3D scanner for alignment. Hull-fixed phased array radar uses very high frequency, so the short wavelength can cause a phase difference of the device due to the small positional error. Since the array antenna is fixed with the hull, it has higher accuracy control than the rotary radar for 4 array surfaces. The study describes a method of checking the flatness of two radar masts manufactured at a factory, a method of aligning masts in a shipyard, and a method of aligning four array pad mounting surfaces. As a tool for this, a 3D laser scanner and a software-based method for comparing survey results with 3D CAD are used. This paper is meaningful as the first example of installing a four-sided hull-fixed phased array radar on a separate mast from a Korean naval ship and deriving a mechanical alignment method.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.22 no.2
• /
• pp.127-133
• /
• 2021

A recent study of people physically falling focused on analyzing the motions of the falls using a recurrent neural network (RNN) and a deep learning approach to get good results from detecting 2D human poses from a single color image. In this paper, we investigate a detection method for estimating the position of the head and shoulder keypoints and the acceleration of positional change using the skeletal keypoints information extracted using PoseNet from an image obtained with a low-cost 2D RGB camera, increasing the accuracy of judgments about the falls. In particular, we propose a fall detection method based on the characteristics of post-fall posture in the fall motion-analysis method. A public data set was used to extract human skeletal features, and as a result of an experiment to find a feature extraction method that can achieve high classification accuracy, the proposed method showed a 99.8% success rate in detecting falls more effectively than a conventional, primitive skeletal data-use method.