http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
庾盛郁(Sungwook Yu) 대한전기학회 2008 전기학회논문지 Vol.57 No.6
This paper presents an efficient architecture for 2ⁿ-point DCT algorithm. The proposed approach makes use of the fact that, in most DCT applications, the scaling operation in the DCT unit can be eliminated and combined with the scaling operation in the quantizer unit. This important property is efficiently exploited with the CORDIC (COordinate Rotation DIgital Computer) algorithm to produce a regular architecture suitable for VLSI implementation. Although there have been several attempts to exploit CORDIC algorithm in developing DCT architectures, the proposed approach provides the most efficient way for scaled DCT applications by completely eliminating the scale factor compensation.
새로운 이중 색인 사상에 의한 다차원 DFT의 파이프라인 구조 개발
庾盛郁(Sungwook Yu) 대한전기학회 2007 전기학회논문지 Vol.56 No.4
This paper presents a new index mapping method for DFT (Discrete Fourier Transform) and its application to multidimensional DFT. Unlike conventional index mapping methods such as DIT (Decimation in Time) or DIF (Decimation in Frequency) algorithms, the proposed method is based on two levels of decomposition and it can be very efficiently used for implementing multidimensional DFT as well as l-dimensional DFT. The proposed pipelined architecture for multidimensional DFT is very flexible so that it can lead to the best tradeoff between performance and hardware requirements. Also, it can be easily extended to higher dimensional DFTs since the number of CEs (Computational Elements) and DCs (Delay Commutators) increase only linearly with the dimension. Various implementation options based on different radices and different pipelining depths will be presented.
SHVC All Intra 공간 스케일러빌리티를 위한 효율적인 모드 결정 알고리즘
신성윤(Sungyoon Shin),유성욱(Sungwook Yu) 대한전기학회 2020 전기학회논문지 Vol.69 No.4
This paper proposes an efficient mode decision method for All Intra (AI) spatial scalability in SHVC. The proposed method can significantly reduce the encoding time by effectively reducing the number of candidate modes in the rough mode decision (RMD) step as well as the number of candidate modes in the rate-distortion optimization (RDO) step. In the RMD step, a new absolute difference of average (ADA) measure is proposed that can dramatically reduce the number of intra-modes to be examined. The proposed method uses different threshold values depending both on the direction and the CU sizes for the best tradeoff between the performance and the encoding complexity. In the RDO step, the proposed method makes use of three kinds of candidate modes, which are base-layer (BL) correlated modes, spatially correlated modes, and the candidate modes from the RMD step. By efficiently combining different kinds of candidate modes to be examined, the proposed method not only reduces the encoding complexity significantly but also shows better performance compared to other mode decision methods.
Sparse estimation residual attention network을 이용한 이미지 denoising network
이지영(JiYoung Lee),유성욱(Sungwook Yu) 한국통신학회 2022 한국통신학회 학술대회논문집 Vol.2022 No.2
본 논문은 기존의 SOTA image denoising model보다 우수한 성능을 나타내는 SERANet모델을 제안한다. 해당 모델은 deep CNN에서 발생할 수 있는 문제점을 보완하기 위해 residual encoder decoder (REDNet)를 기반으로 sparse estimation과 residual attention을 수행한다. sparse estimation은 dilated convolution과 일반적인 convolution layer로 구성되어 있어 computation cost에 대한 부담을 적게 하되 이미지 내의 다양한 정보를 습득할 수 있도록 해 준다. residual attention은 REDNet의 encoder 입력단과 decoder의 출력단의 중간 layer에 위치하며, 채널 내 feature의 값을 추출하여 REDNet이 이미지 디노이즈를 효과적으로 수행할 수 있게 한다. 마지막으로 skip connection으로 연결되어 있는 REDNet을 통해 이미지의 디노이즈를 본격적으로 수행할 수 있다. 결과적으로, 제안하는 모델은 다양한 노이즈 레벨의 AWGN에 대해 SOTA 모델보다 우수한 PSNR 성능을 보인다.