News & Politics

Cache-based motion estimation architecture for real-time HDTV Encoding with H.264/AVC

Description
This paper describes an innovative, pipelined, cache-based architecture for a motion estimation coprocessor based on a predictive/recursive algorithm whose computational complexity is low and independent from the search window. The algorithm and the
Published
of 102
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  STJ OURNALOF S YSTEM R ESEARCH - NUMBER 0 1 FOREWORD P . 1A RT . 1 A High Quality and Search Window Independent Motion Estimation Algorithm and Architecture for MPEG2 Encoding by F. Rovati, D. Pau, E. Piccinelli, L. Pezzoni, J.M. Bard (STMicroelectronics)  P . 4A RT . 2 An Innovative Adaptive Noise Reduction Filter for Moving Pictures Based on Modified Duncan Range Test by A. Borneo, L. Salinari, D. Sirtori (STMicroelectronics)  P . 14A RT . 3 The Zipper-DMT VDSL System: state of the art of the Zipper-VDSL Technology,ST Product Chipset, Integration Strategy and Architecture Evolution by C. Del Toso, M. Combe (STMicroelectronics)  P . 23A RT . 4 Digital still cameras and mobile agents: How to create a distributed service for image processing by M. Mancuso, M. Guarnera (STMicroelectronics), A. Puliafito, M. Villari (University of Messina)  P . 38A RT . 5 A Power Modeling and Estimation Framework for VLIW-based Embedded Systems by L. Benini, D. Bruni (Università degli Studi di Bologna), M. Chinosi, R. Zafalon (STMicroelectronics),C. Silvano (Università degli Studi di Milano), V. Zaccaria (Politecnico di Milano)  P . 52A RT . 6 Superresolution from Low Resolution Videos by Li Hong (STMicroelectronics)  P . 61A RT . 7 The Use of Hyperlines in Light Fields by G. Chen, L. Hong, K. Ng, P. McGuinness (STMicroelectronics), C. Hofsetz, Y. Liu, N. Max (Center for Image Processing and Integrated Computing, U.C. Davis)  P . 74A RT . 8 Hardware implementation of the AES S-box: a case study by M. Macchetti, Alari (Università della Svizzera Italiana), G. Bertoni (Politecnico di Milano)  P . 84A RT . 9 Person identification technique using human iris recognition by C.L. Tisse, L. Martin (STMicroelectronics), L. Torres, M. Robert (Université de Montpellier)  P . 92  2 C OPYRIGHT © STM ICROELECTRONICS , 2003 ST J OURNAL OF S YSTEM R ESEARCH - NUMBER 0 ST J OURNALOF S YSTEM R ESEARCH E DITORIN C HIEF Andrea Cuomo M ANAGING E DITORS Osvaldo M. ColavinFlavio Lorenzelli S CIENTIFIC C OMMITTEE Giulio CasagrandeBruno MurariRobert “Bob” KryziakGuy LauvergeonJefferson Owen E DITORIAL C OMMITTEE Roberto SanninoDenis MestdaghMassimo MancusoRoberto ZafalonGeorge-Qian ChenJean Nicolai A SSOCIATE E DITOR Clara Colombost.journal@st.com D IRETTORE R ESPONSABILE Maria Teresa Gatti R EDAZIONE STMicroelectronics SrlVia C. Olivetti 220041 Agrate Brianza (Milano) - Italy L AYOUTANDPRODUCTION La Carta StampataMilano, Italy P HOTOS STMicroelectronics - ZerodueMilano, Italy S TAMPA C&M Print SasPioltello (Milano), Italy, via Roma 4Iscrizione presso il Tribunale di MonzaNr. 1650/2003 EDITOR IN CHIEF Andrea CuomoCorporate Vice President, STMicroelectronicsAST General Manager ADVISORYBOARD Igor AlexanderImperial College of Science Technology and MedicineGiovanni De Micheli - Stanford UniversityJoseph A.Fisher - Hewlett PackardNick McKeown - Stanford UniversityAlberto Sangiovanni Vincentelli - Berkeley UniversityRobert M.White - CMU MANAGING EDITORS Osvaldo ColavinFlavio Lorenzelli ASSOCIATE EDITOR Clara Colombo  Several years ago , when wedecided to create this journal, we hada clear vision.We didn’t want a “look-mom-how-good-we-are” magazine; we didn’twant an advertising tool; we didn’twant an academic exercise. Wewanted to fulfill a dream we’d hadsince the foundation of AdvancedSystem Technology: to develop world-class knowledge and share it insideand outside ST, while attractingcontributions from partners, institutions,academia.This journal is a first effort in thisdirection. As any newborn, it is still * To the starts through difficulties ST J OURNAL OF S YSTEM R ESEARCH - NUMBER 0 3 WELCOME to the first publicissue of the STJournal of System Research (STJSR). TheSTJSR has been in existence as an internal publication for threeyears, and we are pleased and proud to open it to a widerreadership.As a rule, STJSR issues are thematic. However, to pique yourinterest we have gathered here a collection of papers from theseven previous internal issues, whose guest editors deserve amarked acknowledgment for their invaluable contribution:Roberto Sannino, Denis Mestdagh, Massimo Mancuso, RobertoZafalon, George Chen, and Jean Nicolai.Our collection opens with two papers drawn from the very firstissue of the journal, dedicated to Imaging and Video. Bothcontributions testify to the significant efforts in research aimedat improving the performance of MPEG -2 encoders, from botharchitectural (F. Rovati et al.) and algorithmic (A. Borneo et al.)points of view.Representing the VDSLissue are C. Del Toso and M. Combe,who introduce the integrated Zipper-VDSLchipset developedby STMicroelectronics.The paper by M. Mancuso et al., appearing in the STJSR issueon Image Processing for Digital Still Camera, details anarchitecture for the distributed processing of high-qualityimages, based on the mobile agent communication paradigm.The collection continues with a contribution in the area of System Design for Low Power. The article by L. Benini et al.depicts a methodology for the power consumption estimation of embedded VLIW(Very Long Instruction Word) architectures,such as STMicroelectronics’Lx embedded processors.Two papers, one each by L. Hong and G. Chen et al., introducealgorithms developed in the field of Visual Interface. The formerdescribes a technique that attempts to reconstruct a higherresolution image from a low-resolution input video. The latterdiscusses the concept of hyperlines in light fields and their usein light field rendering.The theme of our most recent issue was Security. M. Marchettiand G. Bertoni, and C.L. Tisse present papers dealing withproblems as disparate as encryption (ASIC implementation of the AES Rijndael algorithm) and biometrics (algorithms forhuman iris recognition).Our ultimate aspiration is that you will learn something fromreading these pages. We look forward to having the privilege of your interest in the future. Osvaldo Colavin, Flavio LorenzelliSTJSR co-editors largely unprepared for the expectationsits parents have for him. But it is willingto learn,to grow fast, andto outperform even the highestexpectations.But no one can grow or learn alone.Therefore we count on all of you tocriticize, correct, and contribute.A learning journey requires teachersand friends: and this is what we askall of you to be. Our success is also yours, and we want to offer anddevelop a forum where new ideas areexposed, exchanged, and challenged.To everyone I wish a good journey. Per aspera ad astra *. Andrea Cuomo Corporate Vice PresidentGeneral Manager,Advanced System TechnologySTMicroelectronics  STJ OURNALOF S YSTEM R ESEARCH - NUMBER 0 1. OVERVIEWON MOTIONESTIMATION 1.1 Motion estimation in MPEG-2 The MPEG-2 standard [1] defines the syntax and semantics of the transmitted bit-stream and the functionality of the decoder.However the encoder is not strictly standardized; any encoderthat produces a valid MPEG-2 bitstream is acceptable. Oneexample encoder can be found in [2].One of the subtasks of MPEG-2 encoding is motion estimation,which we use to compute similarities among successive picturesin order to remove temporal redundancy, i.e., to transmit onlythe difference between successive pictures. In particular, BlockMatching Motion Estimation (BM-ME) is a common way toextract the similarities among pictures, and it is the techniqueselected by the MPEG-2 standard.Fig. 1 shows the basic concepts of BM-ME. We basicallyestimate in one go a “chunk” of the picture. We start by slicingthe current picture in 16 by 16 pixels arrays called“Macroblocks” (a); we then project the position of eachreference macroblock into the prediction frame (b). We definea search window area (c) where we constrain the search for thebest candidate for prediction. We then go on testing the‘similarities’of possible candidate predictor (d) with thecurrent MB (a). Each predictor (d) is uniquely identified by its Fabrizio S. RovatiDanilo PauEmiliano PiccinelliLuca PezzoniJean-Michel BardSTMicroelectronics AH IGH Q UALITYAND S EARCH W INDOW I NDEPENDENT M OTION E STIMATION A LGORITHMAND A RCHITECTUREFOR MPEG -2 E NCODING This paper describes an innovative,pipelined, cache-based architecturefor a motion estimation coprocessorbased on a predictive/recursivealgorithm whose computationalcomplexity is low and not dependenton the search window.The synergies between architectureand algorithmic features allow a highpicture quality, low-area, low-bandwidth,unlimited search window implementationfor standard resolution (PAL/NTSC),low resolution (CIF/QCIF) or highresolution (ATSC) sequences. 4 C OPYRIGHT © IEEE, 2000 R EPRINTED , WITHPERMISSION , FROM A N I NNOVATIVE , H IGH Q UALITYAND S EARCH W INDOW I NDEPENDENT M OTION E STIMATION A LGORITHMAND A RCHITECTUREFOR MPEG-2 E NCONDING , BY F ABRIZIO R OVATI , D ANILO P AU , E MILIANO P ICCINELLI , L UCA P EZZONIAND J EAN -M ICHEL B ARD ,IEEE T RANSACTIONSON C ONSUMER E LECTRONICS , A UGUST 2000, V OL . 46, I SSUE 3, PAGES 697 - 705
Search
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks
SAVE OUR EARTH

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!

x