• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.12 no.6
• /
• pp.567-572
• /
• 2019

In this paper, we propose an algorithm to perform convolution, pooling, and ReLU operations in CNN using binary image and binary kernel. It decomposes 256 gray-scale images into 8 bit planes and uses a binary kernel consisting of -1 and 1. The convolution operation of binary image and binary kernel is performed by addition and subtraction. Logically, it is a binary operation algorithm using the XNOR and comparator. ReLU and pooling operations are performed by using XNOR and OR logic operations, respectively. Through the experiments to verify the usefulness of the proposed algorithm, We confirm that the CNN operation can be performed by converting it to binary logic operation. It is an algorithm that can implement deep running even in a system with weak computing power. It can be applied to a variety of embedded systems such as smart phones, intelligent CCTV, IoT system, and autonomous car.

• Journal of Broadcast Engineering
• /
• v.22 no.2
• /
• pp.182-192
• /
• 2017

In this paper, we propose hand tracking system combined with color and motion information. Most of hand detection and tracking systems are performed by modeling skin color. However, in this approach, since it is highly influenced by light or surrounding objects, accurate values cannot be derived constantly. Also, depending on the skin color, hand tracking may be interrupted by not only the hand but also the background with a color similar to that of the face and skin. Therefore, we design the hand tracking that can effectively track a hand by using motion history image(MHI) and combining it with CamShift. The proposed system is implemented based on C/C++, and the experiments proved that the proposed method shows stable and excellent performance.

• Publications of The Korean Astronomical Society
• /
• v.28 no.1
• /
• pp.7-13
• /
• 2013

We present a multi-dimensional reduction method of the surveyed cube database obtained using a single- dish radio telescope in Taeduk Radio Astronomy Observatory (TRAO). The multibeam receiver system installed at the 14 m telescope in TRAO was not optimized at the initial stage, though it became more stabilized in the following season. We conducted a Galactic Plane survey using the multibeam receiver system. We show that the noise level of the first part of the survey was higher than expected, and a special reduction process seemed to be definitely required. Along with a brief review of classical methods, a multi-dimensional method of reduction is introduced; It is found that the 'background' task within IRAF (Image Reduction and Analysis Facility) can be applied to all three directions of the cube database. Various statistics of reduction results is tested using several IRAF tasks. The rms value of raw survey data is 0.241 K, and after primitive baseline subtraction and elimination of bad channel sections, the rms value turned out to be 0.210 K. After the one-dimensional reduction using 'background' task, the rms value is estimated to be 0.176 K. The average rms of the final reduced image is 0.137 K. Thus, the image quality is found to be improved about 43% using the new reduction method.

• Journal of the Korea Society of Computer and Information
• /
• v.14 no.11
• /
• pp.179-186
• /
• 2009

In this paper, we propose a head detection method based on vertical and horizontal pixel histogram analysis in order to overcome drawbacks of the previous head detection approach using Haar-like feature-based face detection. In the proposed method, we create the vertical and horizontal foreground pixel histogram images from the background subtraction image, which represent the number of foreground pixels in the same vertical or horizontal position. Then we extract feature points of a head region by applying Harris corner detection method to the foreground pixel histogram images and by analyzing corner points. The proposal method shows robust head detection results even in the face image covering forelock by hairs or the back view image in which the previous approaches cannot detect the head regions.

• Journal of the Institute of Convergence Signal Processing
• /
• v.23 no.2
• /
• pp.107-114
• /
• 2022

In order to classify 3D film images as right or wrong, it is necessary to detect the pattern in a 3D film image. However, if the contrast of the pixels in the 3D film image is low, it is not easy to classify as the right and wrong 3D film images because the pattern in the image might not be clear. In this paper, we propose a method of classifying 3D film images as right or wrong by comparing the width at a specific frequency of each histogram after obtaining the histogram. Since, it is classified using the width of the histogram, the analysis process is not complicated. From the experiment, the histograms of right and wrong 3D film images were distinctly different, and the proposed algorithm reflects these features, and showed that all 3D film images were accurately classified at a specific frequency of the histogram. The performance of the proposed algorithm was verified to be the best through the comparison test with the other methods such as image subtraction, otsu thresholding, canny edge detection, morphological geodesic active contour, and support vector machines, and it was shown that excellent classification accuracy could be obtained without detecting the patterns in 3D film images.

• Journal of Korean Neurosurgical Society
• /
• v.61 no.2
• /
• pp.167-179
• /
• 2018

The prevalence of patients with adult spinal deformity (ASD) has been reported as high as 68%. ASD often leads to significant pain and disability. Recent emphasis has been placed on sagittal plane balance and restoring normal sagittal alignment with regards to the three dimensional deformity of ASD. Optimal sagittal alignment has been known to increase spinal biomechanical efficiency, reduce energy expenditure by maintaining a stable posture with improved load absorption, influence better bony union, and help to decelerate adjacent segment deterioration. Increasingly positive sagittal imbalance has been shown to correlate with poor functional outcome and poor self-image along with poor psychological function. Compensatory mechanisms attempt to maintain sagittal balance through pelvic rotation, alterations in lumbar lordosis as well as knee and ankle flexion at the cost of increased energy expenditure. Restoring normal spinopelvic alignment is paramount to the treatment of complex spinal deformity with sagittal imbalance. Posterior osteotomies including posterior column osteotomies, pedicle subtraction osteotomies, and posterior vertebral column resection, as well anterior column support are well known to improve sagittal alignment. Understanding of whole spinal alignment and dynamics of spinopelvic alignment is essential to restore sagittal balance while minimizing the risk of developing sagittal decompensation after surgical intervention.

• Publications of The Korean Astronomical Society
• /
• v.32 no.1
• /
• pp.25-27
• /
• 2017

We are creating all-sky diffuse maps from the AKARI mid-infrared survey data with the two photometric bands centered at wavelengths of 9 and $18{\mu}m$. The AKARI mid-infrared diffuse maps achieve higher spatial resolution and higher sensitivity than the IRAS maps. In particular, the $9{\mu}m$ data are unique resources as an all-sky tracer of the emission of polycyclic aromatic hydrocarbons (PAHs). However, the original data suffer many artifacts. Thus, we have been developing correction methods. Among them, we have recently improved correction methods for the non-linearity and the reset anomaly of the detector response. These corrections successfully reduce the artifact level down to $0.1MJy\;sr^{-1}$ on average, which is essential for discussion on faint extended emission (e.g., the Galactic PAH emission). We have also made progress in the subtraction of the scattered light caused in the camera optics. We plan to release the improved diffuse maps to the public within a year.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.9 no.10
• /
• pp.1095-1102
• /
• 2014

In this paper, We proposed the approach of GPU and motion vector to analysis the Human activity in real-time surveillance system. The most important part, that is detect blob(human) in the foreground. We use to detect Adaptive Gaussian Mixture, Weighted subtraction image for salient motion and motion vector. And then, We use motion vector for human activity analysis. In this paper, the activities of human recognize and classified such as meta-classes like this {Active, Inactive}, {Position Moving, Fixed Moving}, {Walking, Running}. We created approximately 300 conditions for the simulation. As a result, We showed a high success rate about 86~98%. The results also showed that the high resolution experiment by the proposed GPU-based method was over 10 times faster than the cpu-based method.

• Journal of Korea Society of Industrial Information Systems
• /
• v.18 no.4
• /
• pp.53-60
• /
• 2013

We propose a method for improving detection speed of small target detection system using SAD algorithm. First, the proposed method deletes clutters using a median filter. Next, it does closing and opening operation using various size of structure elements, and extracts candidate pixels for a target with subtraction operation between the results of closing and opening operation. It finally detects a small target using a gaussian distance function from the candidate pixels. To improve detection speed, it detects a target performing SAD algorithm only for the predicted target areas for next every 7 frames. The proposed method not only enables a real time process because it considers only predicted area but also shows detecting rate of 97%.

• Journal of Korean Neurosurgical Society
• /
• v.42 no.5
• /
• pp.413-415
• /
• 2007

A 26-year-old man was admitted to our department due to intermittent left hemiparesis for 3 months. Magnetic resonance Image showed subacute infarction in the right precentral gyrus. Digital subtraction angiography and magnetic resonance angiography revealed an aneurysmal protrusion at the right middle cerebral artery (MCA) bifurcation. It was difficult to differentiate the aneurysm from the occlusion of the middle trunk of the MCA trifurcation. Brain single photon emission computerized tomography showed a decrease in perfusion in the right posterior frontal lobe without vascular reserve. Therefore, we planned a superficial temporal artery MCA anastomosis with an exploration of the right MCA bifurcation. Intraoperatively, the aneurysmal opacification on preoperative angiography proved to be the proximal stump of the occluded middle trunk of the MCA trifurcation. An aneurysmal protrusion at the MCA bifurcation does not always indicate an aneurysm. In diagnosing protruding vascular lesions at the MCA bifurcation, the possibility of a vascular stump should be considered according to their angioanatomical appearance and the history of the patient.