untitled - PDF 免费下载

流數理 spinner@gate.sinica.edu.tw 95 年 12 04

錄 2 數流 4 參 5 年度 5 數行 11 料立 14 數 15 15 數 16 識 18 料料 27 料 XML..27 料 31 陸 34 34 36 數 40 數 40 數 41 來 47 1

48 數 48 數 53 55 拾參錄 2

年行數數數了數數立數練理數理聯年立 16 類金拓類年立數類歷年來參行數理年度數年度立數年度立蘭料數年度北數年度數料年度略說數理數行 XML 立索料行 1,000 行 30 數數數數料立數路料料立蘭蘭數料蘭料北料北陸北聯行數數料 3

了數行數流年數流來錄數流數參年不數行拓立數行數流流 Standard Operational Procedure SOP 數參流數數來數行行不力行數數流更識數例說數數料北數缩立料蘭料數數蘭來錄數料聯更料數流數料料六參 4

數流數數利識列識料 5

參年度數行量數量數度行數數行數數料數索料 Metadata 錄數料理數類流數了類 Microfilm 狀狀若行復不數理行見 16mm 35mm 度 21 C 度 50% 100~500 年讀不便量數歷列略 1. 年歷 2. 理率易理 3. 省料省 95% 4. 利 100 年 5. 利讀利讀來 6

1 料數量聯 40 年 ~ 92 年 12 357 76 年 ~ 92 年 12 196 67 年 2 ~ 92 年 12 234 35 年 2 ~ 85 年 12 269 料來數老歷臨若留料料數行料來類類不數更數列流數行復復行數行數立讀料數料 Metadata 讀料數數連料數見錄 1. 立行數量便利索路料度料行兩立料錄料 Metadata 料流數度 7

2. 論識數行行行識理了便行行 Metadata 欄便料易若行行列料度錄錄利理錄來利度立數 1. 數料數不不料了 HTML PDF RTF 了便索若立料不路若料不數 TIFF 不 TIFF G4 JPG 85% PDF 說 1 TIFF Tag Image File Format TIFF ALDUS 1986 年立利 Tag TIFF 數 TIFF LZW Lempel-Ziv-Welch Encoding LZW Huffman's Encoding 度 8

不略利 TIFF G4 256 TIFF TIFF G4 300dpi A4 例 50KB 2 JPEG Joint Photographic Experts Group JPEG International Organization for Standardization ISOInternational Telegraph and Telephone Consultative Committee CCITT 立數 Lossy 念利數 Discrete Cosine Transform DCT 料不留率 JPEG 理 JPEG 例利參數來率 5%~15%JPEG 8 連利 JPEG 來度 TIFF 更利路讀 JPEG 率 JPEG 路 3 JPEG2000 JPEG2000 ISO 15444 JPEG the Joint Photographic Experts Group 2000 年 3 JPEG2000 率 JPEG 30% JPEG 量 JPEG2000 不 LuraTech ACD Systems 率理 ACDSee 3.0 JPEG2000 LWF LWF 4 PDF Portable Document Format PDF Adobe Adobe Acrobat PDF 立不讀 9

PDF 不度見 PDF PDF PDF 了 OCR 識 OCR 識料索料欄 PDF 不了了料索讀 PDF 來不了 Microsoft Reader PDF ebook 5 CEB Chinese Electronic Book CEB 北 Apabi Reader 錄不讀 2 量 A4 300DPI 量 TIFF 不不 TIFF LZW 不 TIFF G4 不不 JPEG 不 JPEG 85% JPEG2000 不 PDF 不 2. 數 1 A. A 行 B B. 度度度度行度 300dpi 10

若度度若例度 72dpi 數例離離量 72dpi 1 1 300dpi 4 行不易 2 數不論度亮度降度行數來說數便行數便數量累量來金數不例行料數數料數料行數 TIFF 不 300~600dpi JBIG 150~300dpi GIF 72dpi 數參錄 3 路路數路路來行數路便利度量降度流 4 行數讀列 PDF Apabi Reader 率 11

數行數理識 Optical Character Recognition OCR 料 1 例數行料行數數 JPG PDF 度 300dpi 數度 150dpi 省省力料例數立料 34 445,584 料欄若識索不料 PDF OCR 識行索料例北數料行若料異漏不識識讀例 ASCII BIG-5 料索料量率 99.98% 若年漏識來率不度 OCR 率不例數不利 OCR 來來 12

廉例利料列料列不行數利數行量料料數 Metadata 兩聯數料料 Metadata XML Internet 3 數數索量力識度率不良識率數數若索不識了力 OCR 數率若類識率數不行行行數見 OCR 識率不若理識率不量便索不力 OCR 識理說了 OCR 識 OCR 識參 OCR 量利行行 13

類行 OCR 識類見不行 OCR 識類數數易 OCR 識率年年度不度率 4 數數 OCR 識見數讀讀不度度不不 OCR 不 150dpi TIFF BMP 理力 OCR 理 JPG 識識率行理 150~200dpi TIFF 利 OCR 行利 OCR TIFF G4 150dpi TIFF G4 300dpi JPG TIFF 300dpi 度 OCR 識度行 OCR TIFF G4 150dpi TIFF G4 300dpi 行 OCR 識 5 利 OCR 識利 OCR 識度 TIFF G4 150dpi 度 TIFF G4 300dpi 度 JPG TIFF 300dpi 不 OCR 識 14

料立立料裡不類不料了數料料理不料索料不類數料不若參理料參見料更料類料來料欄聯錄林 Dublin Core DC 欄理異料不更欄欄行 DC 欄料欄錄 15

數留狀數便了狀不易來 1. 讀來 2. 理論來 3. 行 Commission Intornation De'l E' clairage CIE International Commission on Illumination ICI 1976 年 CIE Yxy 數理 CIE Lab IT8 來年來數便 Charge Coupled Device CCD CCD 金 Complementary Metal-Oxide Semiconductor CMOS 行 ICC Profile 連若狀行例 I2S DiGiBook10000RGB 行流參錄 16

例說數若亮度量例若葉更利 1. 列不更行 2. 行不行度 3. 路路數數數量數離力行數數狀降數度不行數不理 180 度 120 度來兩量若量不見行例拉狀 2 17

180 度 120 度 18

識識說識利數識識例 American National Standard Code for Information Interchange ASCII code BIG-5 料索 OCR 識行識度數度 OCR 不不易索類料理更若利 OCR 數不省量易 OCR OCR 力陸北 OCR 1. 力力 UMAX 來 1987 年立力理識丹識丹識理數識力例識類 Chinese Buddhist Electronic Text Association CBETA 行數力更 CBETA 丹 for CBETA OCR 識 19

2. 立 1991 年切識 Handwritten Recognition 1994 年 Advanced Technology Center ATC OCR 識 OCR 1996 年識 OCR 1997 年識 IBM ViaVoice 識 1998 年識 OCR 識識 3. 1998 年立立立了 OCR 理理行領 OCR 識路類良 OCR 行行 OCR 4. 北北立 1992 年 863 領識 TH-OCR 1985 年 TH TsingHua 領力 TH-OCR 識陸率 65% 識識識度 Microsoft Office 2003 20

5. 北北立 1993 年識 863 金 OCR 識了令 OCR 識理識識車識行識列 OCR 1. 識行 OCR 切索句理 2.OCR OCR 識離讀識切數 OCR 理了識識識理略理識理 1 理識理理理切 A. 理說行 OCR 行不不 OCR 識利理 3 識率 21

B. OCR 識理 A 理不欄欄離欄行切來識理 B a. 4 行不理行切切 b. 列行 (a) 例例若 (b) 若數 (c) 連連不 c. OCR 行識度 0.5 度若度識率度理行度量利 OCR 識數例 Book Restorer 行理 22

行理 23

C. 切行落切行識切利來切例若切切若識切行識識切識行切率度 2 識切識行料異識識說參錄 3 理不識率不易令理 OCR 識率來行 A. 識例識更識欄拉數便降識 B. 量識聯 OCR 聯識來識 3. 識例說行 OCR 識 JPEG TIFF JPEG TIFF JPEG TIFF 列 24

JPG TIFF JPG TIFF JPG TIFF OCR 識 25

行 OCR 識行北行力陸行識度列 OCR 丹識 4.5 識 V3.1 TH-OCR 2003 錄行 OCR 識識行 OCR TIFF 度 150dpi 來便 OCR 識率度率識異識行說識率丹異識識行識率不丹行識亂行識識 TXT 識識率丹丹識行識六說識行理便六識 26

OCR OCR 識識率了識度度不拉數識率 7 OCR 識丹識 TH-OCR 識 4.5 V3.1 2003 錄識率 97% 識率 91.5% 識率 97% 識類識率 <50% 不識率 90% 識度 *.pcx/*.tif/*.jpg/*.bmp *.pcx/*.tif/*.jpg/*. bmp/*.eps/*.msp/*. png/*.psd/*.tga/ *.wmf *.txt/*.rtf/*doc./*xls./* slk./*csv./*html *.txt/*.doc/*.xls/*.h tml *.tif/*.bmp/*.pcx/*.f ax/*.jpg *.rtf/*.html/*.txt/ *.jda/*.wps/*.pdf 27

料料料 XML Metadata Metadata 料料 data about data 料料料料不聯 The International Federation of Library Associations and Institutions IFLAMetadata 來路識料 Metadata 念料 Structure Data About Data Metadata 索 full-text indexing 來 Metadata 切數料便料理利料例數料索來說錄 electronic catalogue 料料索料錄料不錄料領料留錄領料來料料料兩類類識料類料識例類料料了例料欄料料 Extensible Markup Language XMLHypertext Markup Language HTML 料料 28

料料度來料錄料料兩料錄理 administration access preservation use of collections 立料料料料 1. 不 2. 來料連料 3. 料理料料 Functionality 不理 Administrative Descriptive Preservation Use Technical 類 Metadata 6 5 6 Metadata 類類例理理 Metadata used in Administrative managing and administering information resources Descriptive Preservation Use Technical 識 Metadata used 料連 to describe or identify information resources 理 Metadata related 狀 to the preservation management of 數 information resources 類 Metadata 錄錄 related to the level and type of use of information resources Metadata 數 Metadata related to how a system function or Metadata behave 29

行不數料料說料料類數數落數料量料利連來料料料料 International Organization for Standardization ISO1986 年了 Standard Generalized Markup Language SGML SGML 料易料料來料 SGML 來 XML XML 1. XML 路料若立更索行識若來料 Extensible Markup Language, XML markup 錄不讀來來念了了 markup language 裡說了理數來 World Wide Web Consortium W3C1998 年 2 了 XML Recommendation 1.0XML 了 SGML XML 了 SGML 不易 30

來 XML 來說 meta-language 句說 XML 來 XML XML 1 不行 2 3 利路 4 力 5 6 XML 度不 7 力 XML 了來料來 XML XML 1 2 度 3 度不 4 semantics XML 來 2. 領 XML 6 路來利路來數更力識便更行料利了理料料數料不利料利更數領 XML 1 NITF News Industry Text Format International Press Telecommunication Council IPTC 2 NewsML News Markup Language 31

不 3 SportsML Sports Markup Language 錄 4 ProgramGuideML Program Guide Markup Language 5 PRISM Publishing Requirements for Industry Standard Metadata IDEAlliance International Digital Enterprise Alliance 錄 6 XMLNews XMLNews.Org NITF 來 7 RSS Really Simple Syndication RSS Netscape Push Web 料 Blog WiKi 8 NRMF News Records Metadata Format 行錄 Metadata 9 UdnML UDN Markup Language 聯聯 10 XinhuaML Xinhua Markup Language 陸 11 CNTF Chinese News Text Format 料料理 Metadata 理類力領識數料理數類不 32

了不 DVD CD-R 列數料利料理料料識識 Metadata 料理良 Metadata 料料來錄索料量省度索料列來料類理不類留料類離料料類理連類理類類了料料欄類類類狀料若更料料更利料料理 33

陸 7 若更率便來連流理行行行參年理立識 http://web.rdec.gov.tw/cisa/ 契例流行例立數 / 料參料數流北數數北行讀讀列數理數狀不數便行兩 8 量理省力力不易理力不易 34

流料來識行數來說力不度行數行 35

行契契行行陸度理度契利契不契數數契數行參 1. 流行讀契例留行數更錄異利利行行行了漏行錄力浪數流 2. 數料料契了列數量數行數行參 36

數錄 / 數讀行錄數流料來數理數流利行數數流練 1. 流數流兩 1 流行 37

2 度不率流了更行利行流 A. 狀率行行率率 B. 錄練力練 C. 流不降率 D. 若行理流數列 A 數 a. 數數 b. 行數 c. 數錄 B a. b. 行數 c. 數錄 d. 若數 C 錄 a. 錄度 b. c. 錄 2. 練說來流率讀理練數行 38

量利練狀流數更行練練 3. 數契不了識數行論數流不論例理度流度狀度 39

數數量理數識路更了流念數料便利來便數更料來數數量數臨不濫數數數數數數數來數數數行路數 9 領料不 (Content Provider)(Content User) 數 (Digital Rights) 數了料 (Metadata) 便索易行數利數零數了行了數來數數理 40

數年來數流數錄狀數數數數數理數利數數來數利數利數理數數金數利說數理 (Digital Rights Management) 數數理 The chain of hardware and software services and technologies confining the use of digital content to authorized use and users and managing any consequences of that use throughout the entire life cycle of the content. DRM is one kind of content protection technology. 數聯數狀都不行理數狀數數利理數理年來了論數利理數念數數行易理利金數數理料數錄識易理利來說數理 41

了數不更不數數理略狀易流數數理列 1. 良數 2. 理錄例數 3. 數理立來不了念了數來利 4. 數理數利兩不立念更數 5. 數理立數來數數理數數理行易數了濫 42

行數理流裡數數數料參數數讀列狀識金料數料數不不數數 Digital Watermarking 數識數行便度若數數見度兩 Visible Watermarking Invisible Watermarking 數聯 2004 13 數量 1. 度不 2. 不便金 3. 4. 數 5. 度度度 43

不不類例 14 1. 識 2. 數料 3. 識數 Fingerprinting 數料 Information Hiding 切錄便數不不力數不數不若金 Public Key Infrastructure 數理來數金不兩 15 1. 金不若金理不易行度來數 2. 金兩不金金金不金數金金料行易讀金料利不 44

料金數 Digital Signature 力更律力若不數若數立立了度理金立度金金 Public Key Infrastructure PKI 裡來來路易 1. 料 2. 識 3. 易料不 4. 律力利 Rights Expression Language REL 數利兩數利理了數數利利見利列 1. XrML extensible Rights Markup Language XrML 數 1994 年 Xerox PARC ContentGuard 樂數 DRM Metadata 理理例數數樂數數 Microsoft Adobe SONY HP Xerox 數理數料數料流 2. ODRL Open Digital Rights Language 來 45

3. EBX Electronic Book exchage EBX NIST Open ebook Forum OEB EBX ebook 利 1 ebook 識例 ISBN 數識 DOI 2 ebook 金讀金金 EBX 讀 3 ebook 讀 ebook 利列讀 4. MPEG Moving Picture Expetrs Group MPEG 1998 年陸 MPEG2 MPEG4 MPEG7 MPEG21 1998 年 MPEG21 了 16 1 數易 2 數 3 4 數 5 數不易降數流 6 數律 7 易 8 數行 5. SDMI Secure Digital Music Initiative SDMI The Secure Digital Music Initiative(SDMI) 錄來數樂不來樂匿利益不 46

來數數識勵數數陸數理若數勵數更數更年來數數數力數易立不理不數數率不數不便 Apple itunes 樂 17 例 itunes 數樂 Apple FairPlay Apple ipod Sony BMG 年 2005 年 11 Mark Russinovich 露樂了數理 Rootkit 不狀度數理來說數理兩數不數不不數不兩來數省 47

數數 1. 1 2 3 4 5 6 2. 1/ 3. 1 類 A. 數 B. 數 2 類 A. B. 8 數數 1. 4. 2. 5. 3. 6. / 類 1. 數 2. 數類 1. 2. 48

數 1. 類 1 數來 A3 若 A3 行更行狀良 2 料行度料類狀良料若不力復 3 良 A4 更力老易落 4 行度 4800dpi 度離力易率不 5 數度 6 理論料 A2 A1 行不便更度 49

7 料行 A1 立不讀度 2. 類 1 數數來量若數量不行數量若率易率略理來 2 數數 CCD CMOS 數 A1 量率不數 9 數 50

數數 51

度 A2 10 度量 A3 600 不 A4 600 不 A3 600 不 A3 600 不 A3 600 不易落 10 ~ 20 3,000~ 6,000 15 20 8~10 A1 4800 不離力 100 數數 A1 A1 不 300 不 1600 不立 / 力度不不 / 不不 / 300~ 350 450~ 600 160~ 350 20~ 40 100~ 150 52

OCR 數說 OCR 量 OCR 11 丹識 4.5 力 $6,600 識 V3.1 $3,990 行行 TH-OCR2003 錄陸 $120,000 行北陸行數數不數行數行數行數例說金不列數 1. 2. 金例 3. 力 = 數量 / 數量 / 數 4. 力 : * 5. 力 = 數量 / 數量 / 數 6. * 7. 力 = 數量 / 數量 / 數 8. * 9. DVD 數金 53

例說列行數略參練力不列量 1. A4 1A3 15 兩 3 丹識 6,600 年 2 識六 3 3 度 300dpi A4 度 2 30 20 量 30*6*20=3,600 4 150,000+30,000*2+6,600 /3 年 /12 =6,016 / = 6,016+30,000*2 /3,600=18 / 2. 行 A1 1 A1 350 兩 3 識 12 年 2 識六 3 3 度 300dpi A1 2 度 40 3600/40=90 20 量 90*6*20=10,800 4 3500,000+30,000*2+120,000 /3 年 /12 =102,222 = 102,222+30,000*2 /10,800=15 / 54

流行數數流念易來數率降索數流練力率行數流行流數度數識 OCR 行參浪力流 1. 數行率 2. 練流 3. 降數 4. 數 OCR 省力流 OCR 識行行北行陸識更數流參來 OCR 識樂見數識識不來率量料索便利數流行連數度異不更參流數來更力不行數 1 55

3~ 2 96 3 OCR 2 4 OCR 3 Recursive projection analysisconnected component detection 5 Metadata B3-4 6 Metadata B2-1 7 1998 8 1998 9 10 6-11 11 李例 12 13 金龍麗 : 例 14 李例 15 金龍 Wrapper-based 16 56

17 57