• 디지털콘텐츠학회 논문지
• /
• 제16권5호
• /
• pp.671-680
• /
• 2015

Joint Bayesian 방법론[1]은 2012년 발표된 이후 최근까지 최고 성능을 보이는 거의 모든 얼굴인식 알고리즘에서 이진 분류를 위해 사용되고 있지만, 지금까지 이를 개선한 알고리즘은 2D-JB[2] 외에 거의 발표되지 않았다. 우리는 본 논문에서 주어진 얼굴 영상과 이를 좌우 반전시킨 미러 영상을 함께 고려함으로써 Joint Bayesian 방법론의 성능을 향상시킬 수 있는 방법론을 제안한다. 일반적인 패턴인식에서 결정함수 값이 결정경계 또는 임계치에 가까운 경우 오류가 발생할 확률이 높다. 제안한 방법론은 미러 영상의 특징을 이용하여 결정함수 값을 결정경계로부터 멀어지게 함으로써 오류를 줄이는 방법이다. 우리는 LFW DB를 이용한 실험을 통해 제안한 JB 개선 방법론이 기존 JB 방법론보다 1%이상 높은 인식률을 보임을 입증하였다. LFW DB를 이용한 기존 연구들에서 성능을 1% 높이기 위해 많은 학습데이터가 필요했음을 감안할 때, 제안한 방법론은 큰 의미가 있다고 볼 수 있다.

• 한국터널지하공간학회 논문집
• /
• 제20권6호
• /
• pp.989-1002
• /
• 2018

암반 내 그라우팅은 불연속면 내부에 시멘트 그라우트재를 주입하여 주변지반을 강화하는 목적으로 사용된다. 현장에서 다상의 그라우트재의 주입 시 거동특성 및 주입경로인 3차원 절리면의 형태가 사전파악되지 않으므로 정량적인 설계가 어려운 분야중 하나이다. 따라서 현장에서의 그라우트 주입 거동특성을 나타내는 GIN (Grouting Intensity Number) 지표를 이용하여 주입 모니터링을 통해 적절한 시공관리를 수행하는 것이 최적이 방안이다. 본 논문에서는 그라우팅 주입 시 절리면의 거칠기 등급과 물시멘트(W/C)비에 따라 발생하는 압력의 손실을 전산유동해석을 수행하여 조사하였다. 절리면이 거칠수록 그리고 물시멘트비가 높을수록 주입 시 마찰저항은 크게 발생하였으며 해당 결과를 각 조건별 상관식으로 정리하였다.

• Imaging Science in Dentistry
• /
• 제51권2호
• /
• pp.149-154
• /
• 2021

Purpose: This study aimed to investigate the computed tomography and magnetic resonance imaging features of giant cell tumors in the temporomandibular joint region to facilitate accurate diagnoses. Materials and Methods: From October 2007 to June 2020, 6 patients (2 men and 4 women) at Yonsei University Dental Hospital had histopathologically proven giant cell tumors in the temporomandibular joint. Their computed tomography and magnetic resonance imaging findings were reviewed retrospectively, and the cases were classified into 3 types based on the tumor center and growth pattern observed on the radiologic findings. Results: The age of the 6 patients ranged from 25 to 53 years. Trismus was found in 5 of the 6 cases. One case recurred. The mean size of the tumors, defined based on their greatest diameter, was 32 mm (range, 15-41 mm). The characteristic features of all cases were a heterogeneously-enhancing tumorous mass with a lobulated margin on computed tomographic images and internal multiplicity of signal intensity on T2-weighted magnetic resonance images. According to the site of origin, 3 tumors were bone-centered, 2 were soft tissue-centered, and 1 was peri-articular. Conclusion: Computed tomography and magnetic resonance imaging yielded a tripartite classification of giant cell tumors of the temporomandibular joint according to their location on imaging. This study could help clinicians in the differential diagnosis of giant cell tumors and assist in proper treatment planning for tumorous diseases of the temporomandibular joint.

• Journal of the Korean Association of Oral and Maxillofacial Surgeons
• /
• 제26권4호
• /
• pp.373-382
• /
• 2000

Temporomandibular disorders arised from joint itself was caused by inflammation or mechanical irritation of the retrodiskal tissues. Histologic changes of the retrodiskal tissues of temporomandibular joint(TMJ) internal derangement, such as inflammatory cell infiltration, hyalinization, myxoid change, fatty change, arterial obliteration, and so on, were reported, but relationships between magnetic resonance imaging(MRI) findings and histologic findings of the retrodiskal tissue were few reported. The purpose of this study was to find histologic changes of the retrodiskal tissues for status of joint and joint effusion in MRI of the temporomandibular joint which had surgical treatment. Materials were surgically treated 52 temporomandibular joints with internal derangement or osteoarthritis in TMJ clinic, Yongdong Severance Hospital. All joints were confirmed by magnetic resonance T1- and T2-weighted imagings bofere surgery. Status of joint was categorized by normal position, disc displacement with reduction, disc displacement without reduction, disc displacement without reduction associated with osteoarthrosis using preoperative MR T1-weighted images. Magnetic resonance evidence of joint effusion was observed in T2-weighted images and classified by its amount; degree 0(not observed), degree 1(small amount), degree 2(moderate amount), degree 3(large amount). Histologic features were observed whether the retrodiskal tissue has inflammatory cell infiltration, myxoid change, hyalinization, chondroid metaplasia and arterial obliteration. The distribution of elastic fibers were also observed. The results were as follows; 1. Inflammatory cell infiltration was not observed in any retrodiskal tissues. 2. MRI findings such as status of joint and evidence of joint effusion did not have significant relationship with myxoid change, hyalinization, chondroid metaplasia, arterial obliteration of the retrodiskal tissues. 3. Hyalinization and chondroid metaplasia were found in 8 joints(15.4%) and 5 joints(9.6%). All of them were found in disc displacement without reduction and disc displacement without reduction associated with osteoarthrosis. 4. Arterial obliteration was observed more frequently in disc displacement without reduction(55.6%) than disc displacement without reduction associated with osteoarthrosis(28.6%). Further studies with proper controls and more materials will be necessary.

• Journal of Information Processing Systems
• /
• 제12권3호
• /
• pp.381-391
• /
• 2016

The Joint Bayesian (JB) method has been used in most state-of-the-art methods for face verification. However, since the publication of the original JB method in 2012, no improved verification method has been proposed. A lot of studies on face verification have been focused on extracting good features to improve the performance in the challenging Labeled Faces in the Wild (LFW) database. In this paper, we propose an improved version of the JB method, called the two-dimensional Joint Bayesian (2D-JB) method. It is very simple but effective in both the training and test phases. We separated two symmetric terms from the three terms of the JB log likelihood ratio function. Using the two terms as a two-dimensional vector, we learned a decision line to classify same and not-same cases. Our experimental results show that the proposed 2D-JB method significantly outperforms the original JB method by more than 1% in the LFW database.

• 한국치위생학회지
• /
• 제14권3호
• /
• pp.303-310
• /
• 2014

Objectives : The purpose of the study is to investigate the effects of bite force by gender and physical features. Methods : A self-reported questionnaire was filled out by 212 college students in Chungnam province from October to December, 2013. The questionnaire consisted of oral health condition. The bite force of the first molar teeth was measured. Results : Males showed the greater bite force than females(p<0.001). Those who had temporomandibular joint(TMJ) disorder suffered from mouth opening difficulty and weaker bite force than those who had not(p<0.05). Those who had malocclusion showed weaker bite force (p<0.01). Those who had a habit of chewing gums tended to have greater bite force than those who had not(p<0.05). The bite force was correlated with height(r=0.309), weight(r=0.345), and BMI(r=0.249)(p<0.01). Conclusions : Males showed greater bite force than females. The temporomandibular joint (TMJ) disorder, subjective malocclusion, and eating habit also affected the bite force.

• 한국강구조학회 논문집
• /
• 제11권1호통권38호
• /
• pp.1-11
• /
• 1999

본 논문에서는 강구조물의 제작 및 시공에서 용접이음부의 고품질을 확보하기 위하여 강구조물 용접이음부 외부결함의 자동검출에 관한 화상처리 알고리즘을 개발한다. 개발 알고리즘은 광학계의 적절한 배치에 의해 얻어지는 4매의 입력화상을 이용하여 기존의 기법에서 검출할 수 없었던 용접이음부 외부결함을 검출할 수 있음을 보인다. 용접 외부결함이 존재하는 시험편을 제작하고 실험을 통하여 개발 알고리즘의 유용성을 확인하였다. 또한 검출된 용접외부결함의 분류 결과를 육안검사 결과와 비교하였다.

• 대한수의학회지
• /
• 제60권4호
• /
• pp.233-236
• /
• 2020

A five-year-old spayed female Labrador retriever presented with right forelimb lameness one day earlier. Radiographs and computed tomography of the right shoulder joint revealed conspicuous enthesophytes, bone cyst, and bone fragment at the intertubercular groove. Magnetic resonance imaging showed a loss of continuity of the biceps tendon and inhomogeneous hyperintense signal of the surrounding soft tissue and moderate synovial fluid on T2- and proton density-weighted images with contrast enhancement on the postcontrast T1-weighted images. The dog was diagnosed with tenosynovitis and biceps tendon rupture with osteoarthritis based on a comprehensive evaluation of the computed tomography and magnetic resonance imaging features.

• Journal of mucopolysaccharidosis and rare diseases
• /
• 제2권1호
• /
• pp.31-31
• /
• 2016

Purpose: To describle clinical features and enzyme activity of Vietnamese patients with Mucolipidosis type II. Methods: Clinical features, laboratory and plasma lysosom enzyme activity by 4 MU-Fluorometric assay was studied from 2014-2015 at the Northern referral center of Pediatrics - National Children's Hospital. Results: 16 cases (7 girls and 9 boys) were diagnosed with I-cell bases on clinical symptoms and enzyme activities studies. Diagnosis age was $5.93{\pm}4.28$ years, onset age was recognised from birth to 4 years (median 1.25) with the feature of joint stiffness and bone deformation. All cases presented with the feature of joint stiffness, chest deformation and kyphoscoliosis; Fifteen cases (93.7%) had coarse facial features. No patients had hepatosplenomegaly on abdominal ultrasound, 5/15 patients had heart valves disease. Enzyme assay showed ${\alpha}$-Hexosaminidase of $1,885.9{\pm}338.7$ (nmol/mg plasma/17 hrs), ${\alpha}$-Iduronate sulfatase of $4,534.8{\pm}1,062.9nmol/mg$ plasma/4 hrs). Conclusion: Mucolipidosis II seriously affected the life of the patients with skeletal deformities, contractures develop in all joints and cardiac involvement.

• 산업경영시스템학회지
• /
• 제46권3호
• /
• pp.161-169
• /
• 2023

Degenerative arthritis is a common joint disease that affects many elderly people and is typically diagnosed through radiography. However, the need for remote diagnosis is increasing because knee pain and walking disorders caused by degenerative arthritis make face-to-face treatment difficult. This study collects three-dimensional joint coordinates in real time using Azure Kinect DK and calculates 6 gait features through visualization and one-way ANOVA verification. The random forest classifier, trained with these characteristics, classified degenerative arthritis with an accuracy of 97.52%, and the model's basis for classification was identified through classification algorithm by features. Overall, this study not only compensated for the shortcomings of existing diagnostic methods, but also constructed a high-accuracy prediction model using statistically verified gait features and provided detailed prediction results.