OCR

[RDecker]

Acrobat Pro does ok OCR, but it's a single language tool. It cannot do mixed text, at least as of v 9. I haven't tried v 10.

Re. The others listed, could someone list the OS for each? Windows? Mac?

[Ken M. Penner]

Hmmm, would you be able to give Abbot-Smith a try?

Here is the result from Anagnostis:

MANUAL GREEK LEXICON 0F
THE NEW TESTAMENT
A, a, aAipa (q.v), t0, itidecl., AfLwia, liie fli'sl leller oT IHe GI~ßeK
alphabet. AB a BUDQeral, a = I, a = 1999. As a prefix, il QI)Pears
19 Have al leasl Iv9 aDII perhaps Ihi'ße dislillcl selises : I, a'- (before
Q vönel, av7 (־iegalill7, as in a'־YVuioT9ç, â-ôixipç. 2. â-, a- Eyirraöiuß,
Îii(licalillg sominuIiiljT and fellotvshîp as in a'-7A9ûç, a'-KipÀtpirôeu)
a'-0£Âip9ç. 3. Ati i7ilBnßiuß force (EB, s. a), as in â-TsviÇu) iç 99iQ9mneç
asBumed (bul '. BOisasq, a v.).
Aap;v (Heb. li-i-JQ), iBdecl. (m Fil, - 0ç), Aaftai (Ex 414, il.) :
LVIs, Ac 748, He 54 711 gi.+
AßaBBuiv (Heb. 117-ik, ZBB1ttie1i07i; EXX, a7Tù)£ia, ooiy m
syis(l9m Lil., oT ih9pia£eoT lheiuincclclea(l: Ib 26' 2823112 -־, ps88!2,
Pr 1511%), inflecl.; in NT, Aba9ZOn, ih9 aB el OP Ihe Abyss: B8 911.+
־ oßaprji. -Iç (41 ß '09ç), 7uiiitOIII luBrjhô ' iriclapb. (MM, '0T, s.'.)
7181 57n'QTe7is077iß : n CO ll\+
־+Aßßö 0־, -â), iDdßsI. (Ar3in. N79,־, sinphnlis RDi'ai Of -m = Hel).
AU,, fQlhe7), use(l m Ihc phrase ־A ' q׳) »ööQ, auöBr (־- Stv8Is 0q
Mit, I.e.) : Mk 143', R0 Bis, Ga 4ß.+
Aßs Ar|vf) (T, Bec. ,Aß'-), -r!ç, r! (sß. x pa), ABiiuiu, a aiiBIHe! in
hiß Anil-liebaDoii : IDk31.+
AßsA (TyH, A-),;, in(lecl. (Hßb. 541-.. i), A5BI (Ge 42-10) : Hß 114

The problem with Anagnostis is that it requires a 2-bit (black and white) file, in TIF, BMP, or GIF format only. I had to export the PDF scan to a monochrome TIF, which was then dithered.
Here's it with Adobe Acrobat X:

- ι •ι t
'
MANUAL GREEK LEXICON OF
ΤΗΕ NEW TESTAMENT
Α
Α, α, άλφα (q.v.), τό, indecl., alJJlιa, the fiΓst letter of the Greek
alphabet. As a nurneral, α' = 1, α, = 1000. As a prefix, it appears
to have at least two and perhaps th1·ee di stinct senses: 1. ά- (before
a vowel, άν-) negative, as in ά-γνωστος-, ά-δικος-. 2. ά-, ά- copulative,
indicating cornrnunity and felloνvship, as in ά-π-λοvς-, ά-κολοvθέω ,
ά-δελcpός. 3. An intensi 'υe force (LS, s. α), as in ά-τε νίζω is sornetirnes
assurned (but v. Boisacq , s.v.).
Άαpώv (Heb. 7;ϊι':7~), indec l. (in FIJ, -wνος), ΑαΨοπ (Εχ 414, al.) :
Lk 1 5 , Ac 7 4 ο, He 54 711 94 .t
Άβα88ώv (Heb. 7;":'1':;,.~ , clestr1.ιction; LXX, άπ-ώλεια, only i n
Wisdorn Lit., of the place of the 1·uined dead: Jb 266 2822 3112, Ps 881 2,
Pr 1511*), indecl.; in ΝΤ, Abadcloπ, the angel of the Abyss: Re 911 .+
* άβαpής, -ές (<β άρο ς), witl~o1.ιt ιυeight; rnetapb. (ΜΜ , VGT, s.v.)
?Wt burdensorne: π Co 119. +
*+ Άββά (Τ, -ίi), indecl. (Ararn. Ν~~ , emphatic fo1·rn of :l~ = H eb.
:l~, jather), used in the phrase Ά. ό πατήρ, Abba, Fatlιer (v. Swete on
Mk, l.c.): Mk 14 36, Ro 815, Ga 46.t
Άβειληvή (Τ, Rec. Άβι-), -1]ς, ή (sc. χώρα), Abileπe, a district in
the Anti-Lebanon: Lk 31.t
,. Αβελ (WH, 'Ά-), ό, indecl. (Heb. ~:;.,:,) , Abel (Ge 4 2·10): He 114
1224 ; αίμα,. Α., Mt 23 35 , Lk 11 51 .+
Άβι&. (Heb. n~~~, ~i1~~~) , ό , indecl. (in FIJ, Άβίας, -α), Abia,
Abijah. 1. Son of Rehoboam (rπ Ki 141): Mt 17• 2. Α priest of the
line of Eleazar (r Cb 243 • 10) : Lk 15• t
Άβιάθαp, ό, indecl. (Heb. ,ζl~=t~), Abiatl~a1· (r Ki 21 1): Mk 226,t
Άβιληvή, v.s. Άβειληνή.
Άβιού8, ό, indecl. (Heb. ,~i1.,~~), Ab ·i1.ιd, Abil~1.ιd: Mt 1 1".1'
Άβpαάμ (Heb. Ci?1=t~), ό, indecl. (in FlJ, 'Άβpαμος-, -ou; ΜΜ,
VGT, s.v.), Abraharn (Ge 175 al.): l\ft 11 , 2 al.
1

Here it is with ABBYY FineReader 10 Professional Edition:

MANUAL GREEK LEXICON OF THE NEW TESTAMENT

A
A, a, αΧφα (q.v.), to, indecl., alpha, the first letter of the Greek alphabet. As a numeral, a = 1, a = 1000. As a prefix, it appears to have at least two and perhaps three distinct senses: 1. d- (before a vowel, άν-) negative, as in α-γνωστ05, ά-δικ05. 2. «1-, a- copulative, indicating community and fellowship, as in ά-ττλονς, ά-κολονθίω, ά-^λφός. 3. An intensive force (LS, s. a), as in ά-τενίζω is sometimes assumed (but v. Boisacq, s.v.).
Άαρώκ (Heb. אסרון), indecl. (in FLT, -ώ™*). Aaron (Ex 4>\ al.):
Lk I5, Ac 740, He 54 7n 94.+
Άβαδδών׳ (Heb. אבדון: , destruction; LXX, άττώλ««, only in
Wisdom Lit., of the place of the ruined dead : Jb 26f> 2822 3112, Ps 8812, Pr 1511*), indecl.; in NT, Abaddon, the angel of the Abyss: Re 9".t ^ ^άβαρή5, -ί'5 (<£«>')״ without weight; metaph. (MM, VGT, s.v.)
*+ ,Αββά (T, -a), indecl (Aram. אבא , emphatic form of אב = Heb. אב » father), used in the phrase "A. 1 ״raW,p, Abba, Father (v. Swete on Mk, I.e.): Mk 1436, Bo 815, Ga 4«.+
A;ttr־s־ ·י־׳^ :׳i!י ׳(sc■ AlAne·a district in
״ΑβεΧ (WH, ״A־), «, indecl. (Heb. Abel (Ge 4«°): He ll4
1224 ; αΐμα"A., Mt 2335, Lk ll51.t
Άβ1ά (Heb. , 6 ,(אב^הז־, indecl (in F1J, 'Αβίας, -a), AUa,
Abijah. 1. Son of Behoboam (111 Ki 141): Mt I7. 2. A priest of the line of Eleazar (1 Ch 24310 ׳): Lk l5.t
׳Αβιάθαρ, b, indecl. (Heb. א£יגר), Abiathar (1 Ki 21ג) : Mk 226.t
ΆβιΧηΐ'ή, V.8. ,Αβαληνψ
Άβιοιίδ, b, indecl. (Heb. ־ΉΓΡΙΚ), Abkid, Abihud: Mt l".t
'Αβραάμ (Heb. אברהם), ό, indecl. (in FIT, Άβραμο5, -״v: MM,
VGT, s.v.), Abraham (Ge 175 al.): Mt I1»2 al.
1

Here is the result from ABBYY PDF Transformer 3.0:

MANUAL GREEK LEXICON OF
THE NEW TESTAMENT
A
Α, α, αλφα (q.v.), τό, indecl., alpha, the first letter of the Greek
alphabet. As a numeral, a = 1, a, = 1000. As a prefix, it appears
to have at least two and perhaps three distinct senses: 1. a- (before
a vowel, dv־) negative, as in α-γνωστ09, α-δικ09. 2. ά-, ά- copulative,
indicating community and fellowship, as in ά-πλοΰς, ά-κολονθεω,
ά ־ δ ε λ φ ο 3 . 5׳ . An intensive force (LS, s. a), as in ά-τενίζω is sometimes
assumed (but v. Boisacq, s.v.).
Ααρών (Heb. אלץ״ון ), indecl. (in F1J, -ωνοή, Aaron (Ex 4 U , al.):
Lk l 5 , Ac 74 0 , He 5* 7 9 4 + . ״
Άβαδδώι ׳ (Heb. אבדון , destruction; LXX, απώλεια, only in
Wisdom Lit., of the place of the ruined dead : Jb 266 2822 311 2 , Ps 8812,
Pr 1511*), indecl.; in NT, Abaddon, the angel of the Abyss: Ee 9״ .+
*άβαρής, ־e ׳s «βάροή, without xoeight; metaph. (MM, VGT, s.v.)
not burdensome: 11 Co 11A+
*·ϊ'Αββά (Τ, -a), indecl. (Aram. א ב א , emphatic form of א ב = Heb.
א ב , father), used in the phrase Ά. ό πατήρ, Abba, Father (v. Swete on
Mk, I.e.): Mk 1436, Eo 815, Ga 46.t
Άβειληηί (Τ, Eec. Άβι-), -7)5, η (sc. χώρα), Abilene, a district in
the Anti-Lebanon : Lk 34. ג
*Αβελ (WH, ״A־), b, indecl. (Heb. ל ב ל ) , Abel (Ge ־4210): He 11*
122 4 ; αίμα"Α., Mt 2335, Lk l l 5 1 . t
Άβιά (Heb. 6 ,( א ב ^ ה י , אביה , indecl. (in F1J, Άβίας, Abijah. 1. Son of Eehoboam (111 Ki 141): Mt l 7 . 2. A priest of the
line of Bleazar (1 Ch 243> 1 0 ) : Lk l 5 . t
Άβιάθαρ, b, indecl. (Heb. אביתר ), Abiathar (1 Ki 211): Mk 22 6 .t
Άβιληη], V.S. Άβαληνή.
Άβιουδ, b, indecl. (Heb. אביהדד ), Abiud, Abihud: Mt l 1 3 . f
Αβραάμ (Heb. אברהם ), b, indecl. (in F1J, "Αβραμος, -ου; MM,
VGT, s.v.), Abraham (Ge 175 al.): Mt 12 ׳ג al.
1

[mapoulos]

The Perseus project published a very detailed document several years ago on OCR for Greek:

http://www.perseus.tufts.edu/publicatio ... eprint.pdf

There are lots of stats and they also detail much of their procedure.

[Peter Pankonin]

Thanks so much Ken,

Wow, ABBYY certainly does a much better job than whatever U of T used (scribe13 is it?) for Souter. At least the Greek looks Greek, but still a lot of manual correction is required. The other factor with the JPEG2000 scans of Souter is that some (the ones I looked at anyway) are pretty light. Increasing the contrast on the scans might help, at least get the English correct.

What puzzles me is why they even bother making epub and mobi versions of Souter and some of the other scanned lexicons that are available on archive.org...

[Peter Pankonin]

Thanks Alex...very interesting paper! My next question was about the state of Open Source OCR. I have tried various tools like gocr and ocrad on mixed text without much success. OCRopus/Tesseract looks very promising.

[Louis L Sorenson]

There are two strands on Textkit talking about OCR:

Re: Look what I found! Greek downloads!! Other languages too http://www.textkit.com/greek-latin-foru ... ocr#p84931

Re: Looking for Classical Greek OCR Software: http://www.textkit.com/greek-latin-foru ... ocr#p81589

I use ReadIris Pro 12 which can do both Greek and English. It needs some training. But it would be nice to be able to force these programs to use a certain unicode range. A lot of time I find it uses the modern Greek vowel with accents (a different Unicode range than Polytonic).

[Jonathan Robie]

I installed Tesseract 3.0 yesterday (this is very easy on Fedora). I have never used OCR software before, I was very impressed by how well it did with both Greek and English texts.

With Funk's grammar, it did a good job on the English if I said the text was English, and a good job on the Greek if I said it was Greek. But I haven't yet figured out how to tell it that the text contains both ...

[Peter Pankonin]

Jonathan: OCRopus, mentioned in the .pdf file to which Alex referred, may help with multi-lingual texts: "OCRopus(tm) is a state-of-the-art document analysis and OCR system, featuring pluggable layout analysis, pluggable character recognition, statistical natural language modeling, and multi-lingual capabilities."

I haven't been as successful with Tesseract so far (I'm using Mandriva 2010.2), which has version 2.04 and no Greek language files. I'll probably have to build version 3 from source in order to use the ell.traineddata file available on google code.

[Jonathan Robie]

I installed Tesseract 3.0 with the Greek training data on Fedora 14. It is a standard package, so I just did this:

$ yum install tesseract tesseract-langpack-ell

I'll try Ocropus - thanks!