DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Lee, Heung-Kyu | - |
dc.contributor.advisor | 이흥규 | - |
dc.contributor.author | Baek, Yun-Ju | - |
dc.contributor.author | 백윤주 | - |
dc.date.accessioned | 2011-12-13T05:24:09Z | - |
dc.date.available | 2011-12-13T05:24:09Z | - |
dc.date.issued | 1997 | - |
dc.identifier.uri | http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=114170&flag=dissertation | - |
dc.identifier.uri | http://hdl.handle.net/10203/33086 | - |
dc.description | 학위논문(박사) - 한국과학기술원 : 전산학과, 1997.2, [ x, 90 p. ] | - |
dc.description.abstract | In the last decade there has been an increasing need for video compression in many multimedia applications such as desktop video conferencing, videophone, digital video recording, video-on-demand, and high-definition television(HDTV). The video coding system based on wavelet transform, namely, {\em wavelet-based video coding system\/} is promising due to its inherent multiresolution structure. In order to implement these video coding system in real time, the special-purpose VLSI is inevitable choice. In this thesis, we discuss efficient VLSI architectures for the wavelet-based video coding system; mainly for {\em wavelet transform} and {\em motion estimation}. We propose an efficient architecture for 1-D wavelet transform. We present a regular output scheduling for 1-D wavelet transform and systematic systolic mapping into linear array. The main features of our architecture include systolic computation, modularity, regularity, and cascadability. While the conventional systolic scheme has a throughput of period 2N, the proposed architecture has a throughput of period N maintaining the systolic structure where N is number of inputs. We also present the minimization of the number of local interconnections for the feedback data link. We propose a parallel systolic architecture for 2-D wavelet transform. We derive the architecture by using new computing scheme called {\em Row-and-then-Column}, and by applying systematic synthesis procedure with regular output scheduling. The proposed architecture consists of row-directional wavelet transform block and column-directional wavelet transform block. Each block is 2-D mesh-like array with regular interconnections. The area cost of the proposed architecture is 4pq MACs(multipliers and adders) that is more efficient than conventional architectures where p is the wavelet filter size and q is the number of parallel inputs. Furthermore, the proposed architecture has high regularity and modularity which brings down the design c... | eng |
dc.language | eng | - |
dc.publisher | 한국과학기술원 | - |
dc.subject | Wavelet transform | - |
dc.subject | Video coding system | - |
dc.subject | VLSI architecture | - |
dc.subject | Motion estimation | - |
dc.subject | 움직임 추정 | - |
dc.subject | 웨이블릿 변환 | - |
dc.subject | 비디오 코딩 시스템 | - |
dc.subject | VLSI 구조 | - |
dc.title | VLSI architectures for the wavelet-based video coding system | - |
dc.title.alternative | 웨이블릿 기반 비디오 코딩 시스템을 위한 VLSI 구조 | - |
dc.type | Thesis(Ph.D) | - |
dc.identifier.CNRN | 114170/325007 | - |
dc.description.department | 한국과학기술원 : 전산학과, | - |
dc.identifier.uid | 000925159 | - |
dc.contributor.localauthor | Lee, Heung-Kyu | - |
dc.contributor.localauthor | 이흥규 | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.