• Journal of Intelligence and Information Systems
• /
• v.16 no.2
• /
• pp.95-108
• /
• 2010

In this paper, we present a hybrid neural network model for dynamic hand gesture recognition. The model consists of two modules, feature extraction module and pattern classification module. We first propose a modified CNN(convolutional Neural Network) a pattern recognition model for the feature extraction module. Then we introduce a weighted fuzzy min-max(WFMM) neural network for the pattern classification module. The data representation proposed in this research is a spatiotemporal template which is based on the motion information of the target object. To minimize the influence caused by the spatial and temporal variation of the feature points, we extend the receptive field of the CNN model to a three-dimensional structure. We discuss the learning capability of the WFMM neural networks in which the weight concept is added to represent the frequency factor in training pattern set. The model can overcome the performance degradation which may be caused by the hyperbox contraction process of conventional FMM neural networks. From the experimental results of human action recognition and dynamic hand gesture recognition for remote-control electric home appliances, the validity of the proposed models is discussed.

• Progress in Medical Physics
• /
• v.10 no.2
• /
• pp.63-71
• /
• 1999

For the purpose of utilization in 3-D conformal radiotherapy and whole body radiosurgery, the Whole Body 3-Dimensional Topographic Radiation Therapy System has been developed. Whole body frame was constructed in order to be installed on the couch. Radiopaque catheters were engraved on it for the dedicated coordinate system and a MeV-Green immobilizer was used for the patient setup by the help of side panels and plastic rods. By designing and constructing the whole body frame in this way, geometrical limitation to the gantry rotation in 3-D conformal radiotherapy could be minimized and problem which radiation transmission may be altered in particular incident angles was solved. By analyzing CT images containing information of patient setup with respect to the whole body frame, localization and coordination of the target is performed so that patient setup error may be eliminated between simulation and treatment. For the verification of setup, the change of patient positioning is detected and adjusted in order to minimize the setup error by means of comparison of the body outlines using 3 CCTV cameras. To enhance efficiency of treatment procedure, this work can be done in real time by watching the change of patient setup through the monitor. The method of image subtraction in IDL (Interactive Data Language) was used to visualize the change of patient setup. Rotating X-ray system was constructed for detecting target movement due to internal organ motion. Landmark screws were implanted either on the bones around target or inside target, and variation of target location with respect to markers may be visualized in order to minimize internal setup error through the anterior and the lateral image information taken from rotating X-ray system. For CT simulation, simulation software was developed using IDL on GUI(Graphic User Interface) basis for PC and includes functions of graphic handling, editing and data acquisition of images of internal organs as well as target for the preparation of treatment planning.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.49 no.2
• /
• pp.35-44
• /
• 2012

Most recent infrared (IR) sensors have a focal-plane array (FPA) structure. Spatial non-uniformity of a FPA structure, however, introduces unwanted fixed pattern noise (FPN) to images. This non-uniformity correction (NUC) of a FPA can be categorized into target-based and scene-based approaches. In a target-based approach, FPN can be separated by using a uniform target such as a black body. Since the detector response randomly drifts along the time axis, however, several scene-based algorithms on the basis of a video sequence have been proposed. Among those algorithms, the state-of-the-art one based on Kalman filter uses one-directional warping for motion compensation and only compensates for offset non-uniformity of IR camera detectors. The system model using one-directional warping cannot correct the boundary region where a new scene is being introduced in the next video frame. Furthermore, offset-only correction approaches may not completely remove the FPN in images if it is considerably affected by gain non-uniformity. Therefore, for FPN reduction in IR videos, we propose a joint correction algorithm of gain and offset based on bi-directional warping. Experiment results using simulated and real IR videos show that the proposed scheme can provide better performance compared with the state-of-the art in FPN reduction.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.42 no.6
• /
• pp.1-8
• /
• 2005

In this paper, we propose a two-stage head tracking algorithm adequate for real-time active camera system having pan-tilt-zoom functions. In the color convergence stage, we first assume that the shape of a head is an ellipse and its model color histogram is acquired in advance. Then, the min-shift method is applied to roughly estimate a target position by examining the histogram similarity of the model and a candidate ellipse. To reflect the temporal change of object color and enhance the reliability of mean-shift based tracking, the target histogram obtained in the previous frame is considered to update the model histogram. In the updating process, to alleviate error-accumulation due to outliers in the target ellipse of the previous frame, the target histogram in the previous frame is obtained within an ellipse adaptively shrunken on the basis of the model histogram. In addition, to enhance tracking reliability further, we set the initial position closer to the true position by compensating the global motion, which is rapidly estimated on the basis of two 1-D projection datasets. In the subsequent stage, we refine the position and size of the ellipse obtained in the first stage by using shape information. Here, we define a robust shape-similarity function based on the gradient direction. Extensive experimental results proved that the proposed algorithm performs head hacking well, even when a person moves fast, the head size changes drastically, or the background has many clusters and distracting colors. Also, the propose algorithm can perform tracking with the processing speed of about 30 fps on a standard PC.

• The Bulletin of The Korean Astronomical Society
• /
• v.44 no.1
• /
• pp.54.2-54.2
• /
• 2019

Removing noise which occurs inevitably when taking image data has been a big concern. There is a way to raise signal-to-noise ratio and it is regarded as the only way, image stacking. Image stacking is averaging or just adding all pixel values of multiple pictures taken of a specific area. Its performance and reliability are unquestioned, but its weaknesses are also evident. Object with fast proper motion can be vanished, and most of all, it takes too long time. So if we can handle single shot image well and achieve similar performance, we can overcome those weaknesses. Recent developments in deep learning have enabled things that were not possible with former algorithm-based programming. One of the things is generating data with more information from data with less information. As a part of that, we reproduced stacked image from single shot image using a kind of deep learning, conditional generative adversarial network (cGAN). r-band camcol2 south data were used from SDSS Stripe 82 data. From all fields, image data which is stacked with only 22 individual images and, as a pair of stacked image, single pass data which were included in all stacked image were used. All used fields are cut in $128{\times}128$ pixel size, so total number of image is 17930. 14234 pairs of all images were used for training cGAN and 3696 pairs were used for verify the result. As a result, RMS error of pixel values between generated data from the best condition and target data were $7.67{\times}10^{-4}$ compared to original input data, $1.24{\times}10^{-3}$. We also applied to a few test galaxy images and generated images were similar to stacked images qualitatively compared to other de-noising methods. In addition, with photometry, The number count of stacked-cGAN matched sources is larger than that of single pass-stacked one, especially for fainter objects. Also, magnitude completeness became better in fainter objects. With this work, it is possible to observe reliably 1 magnitude fainter object.

• Particle and aerosol research
• /
• v.14 no.1
• /
• pp.17-24
• /
• 2018

Since the motion of the charged particle strongly depends on its charge characteristics, information on charge distributions of target particles is one of the important variables in aerosol research. In this study, charged distribution of atomized NaCl particles were measured using a Gerdien type ion counter. Two kinds of particle charging conditions were used in this study. First, atomized NaCl particles were passed through an aerosol neutralizer to have a Boltzmann charge distribution, and then its charge distribution was measured. In this case, the portion of uncharged particles was compared with the portion obtained from the Boltzmann charge distribution for verifying the suggested experimental method. Second, same experiment was conducted without the aerosol neutralizer to measure the charge distribution of atomized and un-neutralized NaCl particles. In the conclusion, the portion of uncharged, negatively charged and positively charged particles were 19%, 62% and 20%, respectively, for neutralized particles. The atomized particles, which was generated without the aerosol neutralizer, also had almost a zero charge state, but the standard deviation in charge distribution was larger than that of neutralized particles. The test method proposed in this study is expected to be used in various aerosol research fields because it can obtain simple information on the particle charge characteristics more easily and quickly than the existing test methods.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.37 no.6
• /
• pp.10-21
• /
• 2000

The Cellular Noninear Networks structure for Distance Transform(DT) and the robust optical flow detection algorithm based on the DT are proposed. For some applications of optical flows such as target tracking and camera ego-motion computation, correct optical flows at a few feature points are more useful than unreliable one at every pixel point. The proposed algorithm is for detecting the optical flows on the trajectories only of the feature points. The translation lengths and the directions of feature movements are detected on the trajectories of feature points on which Distance Transform Field is developed. The robustness caused from the use of the Distance Transform and the easiness of hardware implementation with local analog circuits are the properties of the proposed structure. To verify the performance of the proposed structure and the algorithm, simulation has been done about various images under different noisy environment.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.6
• /
• pp.31-38
• /
• 2024

This study proposes a novel approach to enhancing the social inclusion and participation of individuals with developmental disabilities. Utilizing cutting-edge virtual reality (VR) technology, we designed and developed a metaverse simulator that enables individuals with developmental disabilities to safely and conveniently experience indoor handicapped handball sports. This simulator provides an environment where individuals with disabilities can experience and practice handball matches. For the modeling and animation of handball players, we employed advanced modeling and motion capture technologies to accurately replicate the movements required in handball matches. Additionally, we ported various training programs, including basic drills, penalty throws, and target games, onto XR (Extended Reality) devices. Through this research, we have explored the development of immersive assistive tools that enable individuals with developmental disabilities to more easily participate in activities that may be challenging in real-life scenarios. This is anticipated to broaden the scope of social participation for individuals with developmental disabilities and enhance their overall quality of life.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.11 no.5
• /
• pp.487-492
• /
• 2018

Recently, various applications for image processing have been ported to the mobile environment due to the expansion of the image shooting on the mobile device. However, in the case of extracting the image foreground, which is one of the most important functions of image synthesis, is difficult since it needs complex calculation. In this paper, we propose an video synthesis technique that can divide images captured by mobile devices into foreground / background and combine them in real time on target images. Considering the characteristics of mobile shooting, our system can extract automatically foreground of input video that contains weak motion when shooting. Using SIMD and GPGPU-based acceleration algorithms, SD-quality images can be processed on mobile in real time.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.10
• /
• pp.139-145
• /
• 2021

In this paper, by using the cattle feed intake, rumination, and in heat monitoring technology, RFI (Residual Feed Intake) monitoring and wearable devices and PCs for predicting abnormalities in budding target web and smart A monitoring system using a phone application was designed and implemented. With the development of this system, the farmer is expected to increase economic efficiency. By analyzing the feed intake, it is possible to identify the difference between the recommended feed amount based on the cow's weight and the feed amount consumed by the cow, and it is expected that early detection of metabolic disorders (abnormality of metabolism) is possible. Farmers using the results of this thesis can distinguish the cows with the most efficient performance, and the 6-axis motion sensor signals input from the wearable device attached to the cow's skin (neck) and the microphone attached to the wearable device. It is possible to measure the cow's rumination and feed intake through the sound of the cow's throat. In the future, improvements will be made to measure additional vital signs such as heart rate and respiration.