• The Journal of Korean Institute of Next Generation Computing
• /
• v.15 no.5
• /
• pp.64-74
• /
• 2019

Although a non-rigid registration has high demands in clinical practice, it has a high computational complexity and it is very difficult for ensuring the accuracy and robustness of registration. This study proposes a method of applying a non-rigid registration to 3D magnetic resonance images of brain in an unsupervised learning environment by using a deep-learning network. A feature vector between two images is produced through the network by receiving both images from two different patients as inputs and it transforms the target image to match the source image by creating a displacement vector field. The network is designed based on a U-Net shape so that feature vectors that consider all global and local differences between two images can be constructed when performing the registration. As a regularization term is added to a loss function, a transformation result similar to that of a real brain movement can be obtained after the application of trilinear interpolation. This method enables a non-rigid registration with a single-pass deformation by only receiving two arbitrary images as inputs through an unsupervised learning. Therefore, it can perform faster than other non-learning-based registration methods that require iterative optimization processes. Our experiment was performed with 3D magnetic resonance images of 50 human brains, and the measurement result of the dice similarity coefficient confirmed an approximately 16% similarity improvement by using our method after the registration. It also showed a similar performance compared with the non-learning-based method, with about 10,000 times speed increase. The proposed method can be used for non-rigid registration of various kinds of medical image data.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.54 no.4
• /
• pp.61-67
• /
• 2017

Low-dose X-ray fluoroscopic image sequences to avoid radiation exposure risk are contaminated by quantum noise. To restore these noisy sequences, we propose a 3D nonlocal means (NLM) filter based on stochastic distancesed can be applied to the denoising of X-ray fluoroscopic image sequences. The stochastic distance is obtained within motion-compensated noise filtering support to remove the Poisson noise. In this paper, motion-adaptive weight which reflected the frame similarity is proposed to restore the noisy sequences without motion artifact. Experimental results including comparisons with conventional algorithms for real X-ray fluoroscopic image sequences show the proposed algorithm has a good performance in both visual and quantitative criteria.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.10 no.3
• /
• pp.357-366
• /
• 1986

The flow characteristics of two-dimensional turbulent offset jet which is discharged parallel to a solid wall has been studied experimentally and numerically. In the experiment, 3-hole pitot tube and 2 channel constant temperature hot-wire anemometer are used to measure local mean velocity, turbulence intensity and Reynolds stress while scannivalve is used to measure the wall pressure distribution. It is confirmed experimentally that local mean velocity is closely related to wall pressure distribution. It is also verified that for large Reynolds numbers and fixed step height there exists a similarity in the distribution of wall pressure coefficient. The maximum values of turbulence intensity occur in the top and bottom mixing layers and the magnitude of Reynolds stress becomes large in the lower mixing layer than in the top mixing layer due to the effect of streamline curvature and entrainment. In the numerical analysis, standard k-.epsilon. model based on eddy viscosity model and Leschziner and Rodi model based on algebraic stress model are adopted. The numerical analyses predict shorter reattachment lengths than the experiment, and this difference is judged to be due mainly to the problem of turbulence model constants and numerical algorithm. This also causes the inconsistency between the two results for other turbulence quantities in the recirculation region and impingement region, which constitutes a subject of a continued future study.

• Journal of Korea Water Resources Association
• /
• v.30 no.6
• /
• pp.611-620
• /
• 1997

The history of Tidal Bench Mark(TBM) at four major tide observation stations on the the Korea West Coast is reviewed. The data concerned with the local mean sea level(LMSL), the datum level(DL), and TBM is collected and checked. The values of LMSL surveyed by Rural Development Corporation(RDC), Office of Hydrographic Affair(OHA), and Office of Port Affair(OPA) are compared so that their unbiased MSLs at four stations are determined. Kriging model is introduced to estimate the design levels for tide; DL, MSL, and high water spring tide(HWOST). The estimated design level is well fitted with the sample data. The value of the identified drifts increase with the latitude. The estimated semi-variograms ${\gamma}(h)$ show self similarity. The values of the ${\gamma}(h)$ for DL and HWOST are 0.005 times as much as the values of ${\gamma}(h)$ of MSL.

• Journal of Institute of Control, Robotics and Systems
• /
• v.7 no.9
• /
• pp.783-789
• /
• 2001

The purpose of this study is to build and to manage environment models with line segments from the sonar range data on obstacles in unknown and varied environments. The proposed method subsequently employs a two-stage data-transform process in order to extract environmental line segments from the range data on obstacles. In the first stage, the occupancy grid extracted from the range data is accumulated to a two-dimensional local histogram grid. In the second stage, a line histogram extracted from an local histogram gird is based on a Hough transform, and matching is a process of comparing each of the segments in the global line segments map against the line segments to detect similarity in overlap, orientation, and arrangement. Each of these tests is made by comparing one of the parameters in the segment representation. After the tests, new line segments are composed to the global line segments map. The proposed technique is illustrated by experiments in an indoor environment.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.9C
• /
• pp.922-929
• /
• 2005

In this paper, we propose the MRS p:Ittern classification techniques by the fusion scheme based on the SpPCA and MLP. A conventional PCA teclulique for the dimension reduction has the problem that it can't find a optimal transformation matrix if the property of input data is nonlinear. To overcome this drawback we extract features by the SpPCA technique which use the local patterns rather than whole patterns. In a next classification step, individual classifier based on MLP calculates the similarity of each class for local features. Finally, MRS patterns is classified by the fusion scheme to effectively combine the individual information. As the simulation results to verify the effectiveness, the proposed method showed more improved classification results than conventional methods.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.9C
• /
• pp.930-938
• /
• 2005

We propose a content-based image retrieval(CBIR) method based on an efncient combination of a color feature and multiresolution texture features. As a color feature, a HSV autocorrelograrn is chosen which is blown to measure spatial correlation of colors well. As texture features, BDIP and BVLC moments are chosen which is hewn to measure local intensity variations well and measure local texture smoothness well, respectively. The texture features are obtained in a wavelet pyramid of the luminance component of a color image. The extracted features are combined for efficient similarity computation by the normalization depending on their dimensions and standard deviation vectors. Experimental results show that the proposed method yielded average $8\%\;and\;11\%$ better performance in precision vs. recall than the method using BDIPBVLC moments and the method using color autocorrelograrn, respectively and yielded at least $10\%$ better performance than the methods using wavelet moments, CSD, color histogram. Specially, the proposed method shows an excellent performance over the other methods in image DBs contained images of various resolutions.

• The KIPS Transactions:PartB
• /
• v.16B no.5
• /
• pp.359-366
• /
• 2009

Analysis of 3-dimensional (3D) protein structure plays an important role of structural bioinformatics. The protein structure alignment is the main subjects of the structural bioinformatics and the most fundamental problem. Protein Structures are flexible and undergo structural changes as part of their function, and most existing protein structure comparison methods treat them as rigid bodies, which may lead to incorrect alignment. We present a new method that carries out the flexible structure alignment by means of finding SSPs(Similar Substructure Pairs) and flexible points of the protein. In order to find SSPs, we encode the coordinates of atoms in the backbone of protein into RDA(Relative Direction Angle) using local similarity of protein structure. We connect the SSPs with Floyd-Warshall algorithm and make compatible SSPs. We compare the two compatible SSPs and find optimal flexible point in the protein. On our well defined performance experiment, 68 benchmark data set is used and our method is better than three widely used methods (DALI, CE, FATCAT) in terms of alignment accuracy.

• Journal of Digital Contents Society
• /
• v.10 no.4
• /
• pp.551-560
• /
• 2009

Under the recent global business environment and active e-commerce era through digital media, the decision on the exporting companies' brand and web domain development is a very crucial strategic choice. Especially, since web domain development is closely related with company's brand strategy, this study is conducted to research the decisive factors of exporting brand strategy on the basis of the need research results on Korean exporting company's overseas brand and domain development strategy. As presented in statistical results, in case of Korean exporting businesses, the level of OEM brand export was when there was high export competitiveness and the level of OEM brand export was high when there was higher uncertainty about the environment. This supports the hypotheses on the basis of existing research documents. However, the similarities of markets failed to bring out any statistical meaningful influence. As the second result, businesses with high export competitiveness tend to select global brand strategy and businesses with high environment uncertainty tend to choose local brand strategy. However, the similarity of markets failed to show any statistical meaningful influence.

• The KIPS Transactions:PartD
• /
• v.12D no.1 s.97
• /
• pp.51-62
• /
• 2005

After figuring out the function of an amino acid sequence, we can infer the function of the other amino acids that have similar sequence composition. Besides, it is possible that we alter protein whose function we know, into useful protein using genetic engineering method. In this process. an original protein amino sequence produces various protein sequences that have different sequence composition. Here, a systematic technique is needed to manage protein version sequences and reference data of those sequences. Thus, in this paper we proposed a technique of managing protein version sequences based on local sequence alignment and a technique of managing protein historical reference data using Trigger This method automatically determines the similarity between an original sequence and each version sequence while the protein version sequences are stored into database. When this technique is employed, the storage space that stores protein sequences is also reduced. After storing the historical information of protein and analyzing the change of protein sequence, we expect that a new useful protein and drug are able to be discovered based on analysis of version sequence.