• International Journal of Fuzzy Logic and Intelligent Systems
• /
• v.14 no.3
• /
• pp.209-215
• /
• 2014

Sequence tagging is the task of predicting frame-wise labels for a given input sequence and has important applications to diverse domains. Conventional methods such as maximum likelihood (ML) learning matches global features in empirical and model distributions, rather than local features, which directly translates into frame-wise prediction errors. Recent probabilistic sequence models such as conditional random fields (CRFs) have achieved great success in a variety of situations. In this paper, we introduce a novel discriminative CRF learning algorithm to minimize local feature mismatches. Unlike overall data fitting originating from global feature matching in ML learning, our approach reduces the total error over all frames in a sequence. We also provide an efficient gradient-based learning method via gradient forward-backward recursion, which requires the same computational complexity as ML learning. For several real-world sequence tagging problems, we empirically demonstrate that the proposed learning algorithm achieves significantly more accurate prediction performance than standard estimators.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2015.11a
• /
• pp.193-196
• /
• 2015

조건부대체 알고리즘(CRA: Conditional Replenishment Algorithm)은 좌우 영상의 해상도가 서로 다른 융합형 3DTV 서비스 환경에서 입체영상의 화질을 개선하기 위해 제안된 기법이다. 조건부대체 알고리즘에서는 비용함수를 도입하여 quad-tree 구조를 가지는 가변크기의 처리단위 (PU: Processing Unit)의 최적 모드를 결정하는데, 본 논문에서는 관심 PU 의 모드를 결정하는 단계에서 주변 PU 들의 모드를 함께 고려함으로써 모드가 고립되는 현상을 방지하는 방법을 제안한다. 제안 방법을 통해 기존의 CRA 의 결과로 발생 가능한 특정 PU의 두드러짐 현상을 제거할 수 있기 때문에 합성된 입체 영상의 주관적 화질이 향상된다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2012.11a
• /
• pp.98-100
• /
• 2012

본 논문에서는 좌우 해상도가 서로 다른 융합형 3DTV 시스템에서 합성된 3D 영상의 화질을 개선하는 방법인 조건부 대체 알고리듬에 적용하기 위한 비트율 제어기법을 제안한다. 조건부 대체 알고리듬에서는 쿼드트리 형태의 가변크기 블록마다 좌우 영상간의 양안시차 정보를 활용하되, 이 정보의 전송을 위해서 부가적인 대역폭이 요구되기 때문에 전송 대역폭의 상황에 따라 부가 정보의 양을 효율적으로 조절할 필요가 있다. 이를 위해 비용함수를 정의하고 가중치를 적용하여 최적의 모드를 결정하는 방법을 제안한다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2011.11a
• /
• pp.4-5
• /
• 2011

본 논문에서는 비대칭적 화질을 갖는 스테레오 3DTV 시스템에서 좌우 두 영상의 화질차이가 합성한 스테레오 3D 영상의 주관적 화질을 떨어뜨릴 정도로 현저한 경우, 이 화질 차이를 보상하기 위하여 조건부 대체를 포함한 양안시차 맵을 이용하는 방법을 제안한다. PSNR값 비교를 통하여 화질 개선 정도를 보았고 0.37~18dB 정도의 PSNR 개선을 확인하였다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2021.06a
• /
• pp.61-62
• /
• 2021

본 논문에서는 조건부 wavenet을 이용한 음성 신호의 잡음 제거 기술을 제안한다. 기존의 음성 신호 잡음 제거 기술은 스펙트로그램을 기반으로 발전되어 왔으나, 잡음으로 인해 변형된 원음의 위상 정보를 복원할 수 없는 문제점을 가진다. 이를 해결하기 위해 시간 영역에서 전 과정을 실행하는 기계학습 모델인 wavenet을 사용하여 음성 신호의 잡음을 제거하는 방법을 제안한다. 특히, 잡음 종류를 조건으로 입력하여 성능 향상을 얻도록 한다. 성능 평가를 통하여 제안 방법이 시간 영역에서 잡음을 감소시킬 수 있음을 확인하였다.

• Journal of the Korean Home Economics Association
• /
• v.48 no.8
• /
• pp.113-128
• /
• 2010

The purposes of this study were 1) to divide consumers into segments according to their consumption attitude and subjective knowledge about natural dyed products 2) to develop a profile for each segment with respect to the media types, the preferred media programs, the information sources, the lifestyle activities and the demographics. Data(n = 213) was collected using questionnaires via the internet. By cluster analysis of the consumers' attitude and their subjective knowledge about natural dyed products, four groups were identified 1) the loyal consumption group (20.1%), 2) the conditional/reluctant consumption group(36.6%), 3) the negative/reluctant consumption group(24.2%) and 4) the non purchasers(19.1%). Several characteristics of these four groups were then compared by ANOVA and chi-square statistics. The results indicated that the consumption attitude and subjective knowledge could be the criteria for segmenting the consumers and that the segmented groups had unique consumer characteristics. Implications of this study were discussed.

• Journal of Korea Multimedia Society
• /
• v.19 no.3
• /
• pp.531-538
• /
• 2016

In this paper, we propose a novel post-processing method of semantic segmentation from indoor scenes with RGBD inputs. For accurate segmentation, various post-processing methods such as superpixel from color edges or Conditional Random Field (CRF) method considering neighborhood connectivity have been used, but these methods are not efficient due to high complexity and computational cost. To solve this problem, we maximize the efficiency of post processing by using depth superpixel extracted from disparity image to handle object silhouette. Our experimental results show reasonable performances compared to previous methods in the post processing of semantic segmentation.

• Smart Media Journal
• /
• v.10 no.4
• /
• pp.28-34
• /
• 2021

Drape is one of the factors that determine the shape of clothes and is one of the very important factors in the textile and fashion industry. At a time when non-face-to-face transactions are being activated due to the impact of the coronavirus, more and more companies are asking for drape value. However, in the case of small and medium-sized enterprises (SMEs), it is difficult to measure the drape, because they feel the burden of time and money for measuring the drape. Therefore, this study aimed to generate a drape image for the material property value input using a conditional adversarial neural network through 3D simulation images generated by measuring digital properties. A drape image was created through the existing 736 digital property values, and this was used for model training. Then, the drape value was calculated for the image samples obtained through the generative model. As a result of comparing the actual drape experimental value and the generated drape value, it was confirmed that the error of the peak number was 0.75, and the average error of the drape value was 7.875

• Journal of Broadcast Engineering
• /
• v.16 no.4
• /
• pp.647-656
• /
• 2011

Motion estimation using one-bit transform and two-bit transform reduces the complexity for computation of matching error; however, the peak signal-to-noise ratio (PSNR) is degraded. Modified 1BT (M1BT) and modified 2BT (M2BT) have been proposed to compensate degraded PSNR by adding conditional local search. However, these algorithms require many additional search points in fast moving sequences with a block size of $16{\times}16$. This paper provides more efficient search method by preparing candidate blocks using the number of non-matching points (NNMP) than the conditional local search. With this NNMP-based search, we can easily obtain candidate blocks with small NNMP and efficiently search final motion vector. Experimental results show that the proposed algorithm not only reduces computational complexity, but also improves PSNR on average compared with conventional search algorithm used in M1BT, M2BT and AM2BT.

• Journal of Broadcast Engineering
• /
• v.17 no.2
• /
• pp.354-362
• /
• 2012

Recently, High Efficiency Video Coding (HEVC) is under development jointly by MPEG and ITU-T for the next international video coding standard. Compared to the previous standards, HEVC supports variety of splitting units, such as coding unit (CU), prediction unit (PU), and transform unit (TU). Among them, it has been known that the recursive quadtree structure of CU can improve the coding efficiency while the encoding complexity is increased significantly. In this paper, a simple conditional probability to predict the early termination condition of recursive unit structure is introduced. The proposed conditional probability is estimated based on Bayes' formula from local statistics of rate-distortion costs in encoder. Experimental results show that the proposed method can reduce the total encoding time by about 32% according to the test configuration while the coding efficiency loss is 0.4%-0.5%. In addition, the encoding time can be reduced by 50% with 0.9% coding efficiency loss when the proposed method was used jointly with HM4.0 early CU termination algorithm.