• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.7
• /
• pp.279-286
• /
• 2021

Recently, We are carrying out a policy of physical distancing of at least 1m from each other to prevent the spreading of COVID-19 disease in public places. In this paper, we propose a method for measuring distances between people in real time and an automation system that recognizes objects that are within 1 meter of each other from stereo images acquired by drones or CCTVs according to the estimated distance. A problem with existing methods used to estimate distances between multiple objects is that they do not obtain three-dimensional information of objects using only one CCTV. his is because three-dimensional information is necessary to measure distances between people when they are right next to each other or overlap in two dimensional image. Furthermore, they use only the Bounding Box information to obtain the exact coordinates of human existence. Therefore, in this paper, to obtain the exact two-dimensional coordinate value in which a person exists, we extract a person's key point to detect the location, convert it to a three-dimensional coordinate value using Stereo Vision and Camera Calibration, and estimate the Euclidean distance between people. As a result of performing an experiment for estimating the accuracy of 3D coordinates and the distance between objects (persons), the average error within 0.098m was shown in the estimation of the distance between multiple people within 1m.

• Journal of the Korean Wood Science and Technology
• /
• v.47 no.2
• /
• pp.229-238
• /
• 2019

Various wood defects occur during tree growing or wood processing. Thus, to use wood practically, it is necessary to objectively assess their quality based on the usage requirement by accurately classifying their defects. However, manual visual grading and species classification may result in differences due to subjective decisions; therefore, computer-vision-based image analysis is required for the objective evaluation of wood quality and the speeding up of wood production. In this study, the SIFT+k-NN and CNN models were used to implement a model that automatically classifies knots and analyze its accuracy. Toward this end, a total of 1,172 knot images in various shapes from five domestic conifers were used for learning and validation. For the SIFT+k-NN model, SIFT technology was used to extract properties from the knot images and k-NN was used for the classification, resulting in the classification with an accuracy of up to 60.53% when k-index was 17. The CNN model comprised 8 convolution layers and 3 hidden layers, and its maximum accuracy was 88.09% after 1205 epoch, which was higher than that of the SIFT+k-NN model. Moreover, if there is a large difference in the number of images by knot types, the SIFT+k-NN tended to show a learning biased toward the knot type with a higher number of images, whereas the CNN model did not show a drastic bias regardless of the difference in the number of images. Therefore, the CNN model showed better performance in knot classification. It is determined that the wood knot classification by the CNN model will show a sufficient accuracy in its practical applicability.

• Korean Journal of Biological Psychiatry
• /
• v.26 no.1
• /
• pp.22-31
• /
• 2019

Objectives Previous studies have revealed inconsistent results on amygdala volume in adult bipolar disorder (BD) patients compared to healthy controls (HC). Since the amygdala encompasses multiple subregions, the subtle volume changes in each amygdala nucleus might have not been fully reflected in the measure of the total amygdala volume, causing discrepant results. Thus, we aimed to investigate volume changes in each amygdala subregion and their association with subtypes of BD, lithium use and clinical status of BD. Methods Fifty-five BD patients and 55 HC underwent T1-weighted structural magnetic resonance imaging. We analyzed volumes of the whole amygdala and each amygdala subregion, including the anterior amygdaloid area, cortico-amygdaloid transition area, basal, lateral, accessory basal, central, cortical, medial and paralaminar nuclei using the atlas in the FreeSurfer. The volume difference was analyzed using a one-way analysis of covariance with individual volumes as dependent variables, and age, sex, and total intracranial volume as covariates. Results The volumes of whole right amygdala and subregions including basal nucleus, accessory basal nucleus, anterior amygdaloid area, and cortico-amygdaloid transition area in the right amygdala of BD patients were significantly smaller for the HC group. No significant volume difference between bipolar I disorder and bipolar II disorder was found after the Bonferroni correction. The trend of larger volume in medial nucleus with lithium treatment was not significant after the Bonferroni correction. No significant correlation between illness duration and amygdala volume, and insignificant negative correlation were found between right central nucleus volume and depression severity. Conclusions Significant volume decrements of the whole amygdala, basal nucleus, accessory basal nucleus, anterior amygdaloid area, and cortico-amygdaloid transition area were found in the right hemisphere in adult BD patients, compared to HC group. We postulate that such volume changes are associated with altered functional activity and connectivity of amygdala nuclei in BD.

• 재활복지
• /
• v.21 no.3
• /
• pp.129-146
• /
• 2017

In an event of an emergency, such as fire in a building, visually impaired and blind people are prone to exposed to a level of danger that is greater than that of normal people, for they cannot be aware of it quickly. Current fire detection methods such as smoke detector is very slow and unreliable because it usually uses chemical sensor based technology to detect fire particles. But by using vision sensor instead, fire can be proven to be detected much faster as we show in our experiments. Previous studies have applied various image processing and machine learning techniques to detect fire, but they usually don't work very well because these techniques require hand-crafted features that do not generalize well to various scenarios. But with the help of recent advancement in the field of deep learning, this research can be conducted to help solve this problem by using deep learning-based object detector that can detect fire using images from security camera. Deep learning based approach can learn features automatically so they can usually generalize well to various scenes. In order to ensure maximum capacity, we applied the latest technologies in the field of computer vision such as YOLO detector in order to solve this task. Considering the trade-off between recall vs. complexity, we introduced two convolutional neural networks with slightly different model's complexity to detect fire at different recall rate. Both models can detect fire at 99% average precision, but one model has 76% recall at 30 FPS while another has 61% recall at 50 FPS. We also compare our model memory consumption with each other and show our models robustness by testing on various real-world scenarios.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.30 no.3
• /
• pp.229-235
• /
• 2019

The implementation of an in-band full-duplex wireless communication system is demonstrated in this study. In the analog/RF domain, the self-interference(SI) signal is reduced using a separate antenna for the transmitter and receiver paths, and most of the SI signal is canceled in the digital domain. A software defined radio(SDR) is used to implement the in-band full-duplex wireless communication system. The USRP X310 device uses transmitting and receiving antennas. By adjusting the gain of the transmitting and receiving ends of the SDR device, the magnitude of the SI signal entering the receiving antenna, and the size of the received signal from the outside, are both set to -64 dB. To verify the in-band full-duplex wireless communication performance, the source data is image and orthogonal frequency-division multiplexing is used for modulation. A WiFi standard frame with a carrier frequency of 2.67 GHz and bandwidth of 20 MHz is used. In the received signal, the SI signal is canceled by digital signal processing and the SI signal is attenuated by up to 34 dB. OFDM demodulation was impossible when the SI signal was not removed. However, the bit error rate is reduced to $2.63{\times}10^{-5}$ when the SI signal is attenuated by 34 dB, and no error is detected in the 100 Mbit data output as a result of passing through the Viterbi decoder.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.36 no.6
• /
• pp.565-572
• /
• 2018

Recently, UAVs (Unmanned Aerial Vehicles) or drones have gained popularity for the engineering surveying and mapping because they enable the rapid data acquisition and processing as well as their operation cost is low. The applicable fields become much wider including the topographic monitoring, agriculture, and forestry. It is reported that the high geospatial accuracy is achievable with the drone photogrammetry for many applications. However most studies reported the best achievable mapping results using well-distributed ground control points though some studies investigated the impact of control points on the accuracy. In this study, we focused on the drone mapping of corridors such as roads and pipelines. The distribution and the number of control points along the corridor were diversified for the accuracy assessment. In addition, the effects of the camera self-calibration and the number of the image strips were also studied. The experimental results showed that the biased distribution of ground control points has more negative impact on the accuracy compared to the density of points. The prior camera calibration was favored than the on-the-fly self-calibration that may produce poor positional accuracy for the case of less or biased control points. In addition, increasing the number of strips along the corridor was not helpful to increase the positional accuracy.

• Journal of Broadcast Engineering
• /
• v.23 no.6
• /
• pp.855-865
• /
• 2018

Sound event detection is one of the research areas to model human auditory cognitive characteristics by recognizing events in an environment with multiple acoustic events and determining the onset and offset time for each event. DCASE, a research group on acoustic scene classification and sound event detection, is proceeding challenges to encourage participation of researchers and to activate sound event detection research. However, the size of the dataset provided by the DCASE Challenge is relatively small compared to ImageNet, which is a representative dataset for visual object recognition, and there are not many open sources for the acoustic dataset. In this study, the sound events that can occur in indoor and outdoor are collected on a larger scale and annotated for dataset construction. Furthermore, to improve the performance of the sound event detection task, we developed a dual CNN structured sound event detection system by adding a supplementary neural network to a convolutional neural network to determine the presence of sound events. Finally, we conducted a comparative experiment with both baseline systems of the DCASE 2016 and 2017.

• Korean Journal of Remote Sensing
• /
• v.34 no.6_3
• /
• pp.1329-1350
• /
• 2018

There have been many applications to observe Earth using synthetic aperture radar (SAR) since it could acquire Earth observation data without reference to weathers or local times. However researches about digital map generation using SAR have hardly been performed due to complex raw data processing. In this study, we suggested feasibility of producing digital map using SAR stereo images. We collected two sets, which include an ascending and a descending orbit acquisitions respectively, of KOMPSAT-5 stereo dataset. In order to suggest the feasibility of digital map generation from SAR stereo images, we performed 1) rational polynomial coefficient transformation from radar geometry, 2) digital resititution using KOMPSAT-5 stereo images, and 3) validation using digital-map-derived reference points and check points. As the results of two models, root mean squared errors of XY and Z direction were less than 1m for each model. We discussed that KOMPSAT-5 stereo image could generated 1:25,000 digital map which meets a standard of the digital map. The proposed results would contribute to generate and update digital maps for inaccessible areas and wherever weather conditions are unstable such as North Korea or Polar region.

• Korean Journal of Remote Sensing
• /
• v.34 no.6_3
• /
• pp.1415-1425
• /
• 2018

Kompsat-5 is the first Earth Observation Satellite which is equipped with an SAR in Korea. SAR images are generated by receiving signals reflected from an object by microwaves emitted from a SAR antenna. Because the wavelengths of microwaves are longer than the size of particles in the atmosphere, it can penetrate clouds and fog, and high-resolution images can be obtained without distinction between day and night. However, there is no color information in SAR images. To overcome these limitations of SAR images, colorization of SAR images using Cycle GAN, a deep learning model developed for domain translation, was conducted. Training of Cycle GAN is unstable due to the unsupervised learning based on unpaired dataset. Therefore, we proposed MS Cycle GAN applying multi-scale discriminator to solve the training instability of Cycle GAN and to improve the performance of colorization in this paper. To compare colorization performance of MS Cycle GAN and Cycle GAN, generated images by both models were compared qualitatively and quantitatively. Training Cycle GAN with multi-scale discriminator shows the losses of generators and discriminators are significantly reduced compared to the conventional Cycle GAN, and we identified that generated images by MS Cycle GAN are well-matched with the characteristics of regions such as leaves, rivers, and land.

• Journal of Biomedical Engineering Research
• /
• v.40 no.3
• /
• pp.105-115
• /
• 2019

Brain tumor surgery may be difficult, but it is also incredibly important. The technological improvements for traditional brain tumor surgeries have always been a focus to improve the precision of surgery and release the potential of the technology in this important area of the body. The need for precision during brain tumor surgery has led to an increase in Robotic-assisted surgeries (RAS). One of the challenges to the widespread acceptance of RAS in the neurosurgery is to recognize invisible tumor accurately. Therefore, it is important to detect brain tumor size and location because surgeon tries to remove as much tumor as possible. In this paper, we proposed brain tumor detection procedures for MRI (Magnetic Resonance Imaging) system. A method of automatic brain tumor detection is needed to accurately target the location of the lesion during brain tumor surgery and to report the location and size of the lesion. In the qualitative assessment, the proposed method showed better results than those obtained with other brain tumor detection methods. Comparisons among all assessment criteria indicated that the proposed method was significantly superior to the threshold method with respect to all assessment criteria. The proposed method was effective for detecting brain tumor.