Sei sulla pagina 1di 9

Optical Character Recognition (OCR)

GIST of OCR Testing

Sl Name of Sub modules Test Font Method of


Font
No Language Data Size Efficiency Improvement
Name
Size
DA DS CR

Sample Major 96% to Spell


1 Hindi Fonts 12 Checker,
Size of to 98 %
√ √ √ Including Algorithm
1000 36
pages Digital Improve-
ones ment,
Exhaustive
DVB- TT Testing
Yogesh 11
To
DVB-TT 32
Surekh

Metal
Type
GIST of OCR Testing

Sl Name of Sub modules Test Font Method of


Font
No Language Data Size Efficiency Improvement
Name
Size
DA DS CR

Not DVB- TT 90% at Spell


2 Marathi Yogesh 11 Checker,
√ √ √ available to word
level Algorithm
DVB-TT 32 Improve-
Surekh ment,
Exhaustive
Metal Testing
Type
Tested on Gurumukhi- Spell Checker,
12-20, Algorithm
97%
3. Punjabi -- -- -- 16 Pages IIYS, 14-24, Improvement,
Foe each CDAC (PN- 12-20,
character TT Amar), 12-20, Exhaustive
fonts and Punjabi, 12-20 Testing
4 font Primaja,
sizes each Anandpur
Sahib
GIST of OCR Testing

Sl Name of Sub modules Test Font Method of


Font
No Language Data Size Efficiency Improvement
Name
Size
DA DS CR

Not DVB- TT 90% at Spell


2 Marathi Yogesh 11 Checker,
available to word
level Algorithm
DVB-TT 32 Improve-
Surekh ment,
Exhaustive
Metal Testing
Type
Tested on Gurumukhi- Spell Checker,
12-20, Algorithm
97%
3. Punjabi 16 Pages IIYS, 14-24, Improvement,
Foe each CDAC (PN- 12-20,
character TT Amar), 12-20, Exhaustive
fonts and Punjabi, 12-20 Testing
4 font Primaja,
sizes each Anandpur
Sahib
GIST of OCR Testing

Sl Name of Sub modules Test Font Method of


Font
No Language Data Size Efficiency Improvement
Name
Size
DA DS CR

Tested on Hemalatha 95 to Spell


4 Telugu -- -- √ 12-24, Checker,
80-90 97%
Pages Algorithm
Harsha- 16-20,
Improve-
-priya
ment,
Exhaustive
SreeLipi
14-20, Testing
Ann font 14
Family

Spell Checker,
Tamil Tested on Algorithm
5. √ -- √ 600 pages Any font
12-36, 98%
Improvement,
Exhaustive
Testing
GIST of OCR Testing

Sl Name of Sub modules Test Font Method of


Font
No Language Data Size Efficiency Improvement
Name
Size
DA DS CR

95 to Spell
Kannada Multi Multi
6 Not 96% Checker,
-- -- -- Font Size
mentioned Algorithm
Improve-
ment,
Exhaustive
Testing
ML-TT 12,14,
Not
Kartika,
Malayalam mentioned 16,18
7. -- -- -- MalBrub Spell Checker,
hmi 12 to
90 to Algorithm
16
Manorma 92% Improvement,
Fonts 12 to Exhaustive
16 Testing
Current
Books 12,14
GIST of OCR Testing

Sl Name of Sub modules Test Font Method of


Font
No Language Data Size Efficiency Improvement
Name
Size
DA DS CR

Sample 1.Modular 93% Spell


8 Oriya -- Shree 14, 16 Checker,
-- √ size of
about 100 Algorithm
pages 2.CDAC Improve-
Font ment,
Exhaustive
Testing

Sample
size of Ratnagiri Spell Checker,
12-36 95%
9. Assamese -- -- √ about Algorithm
100 Improvement,
pages Exhaustive
Testing
GIST of OCR Testing

Sl Name of Sub modules Test Font Method of


Font
No Language Data Size Efficiency Improvement
Name
Size
DA DS CR

Sample Conves 97% Spell


10. Bangla the major 12-36 Checker,
√ √ √ size of
about 100 fonts used Algorithm
pages for Improve-
Publishing ment,
Exhaustive
Testing

DA => Document Analysis


√ => Facility exists
DS => Document Synthesis
-- => Facility not available
CR => Character Recognition Engine
Summary of the Bangla-OCR Test Result performed by STQC

1. Accuracy : Character Level 96.43 %

2. Accuracy : Word Level 90.80 %

3. Speed ( Avg. time for converting 45 Character per second


to .tif or .pc format)
4. Noise Reduction 3.00
(AUT is rated on the scale of 1-5
5 being the best)
5. Skew detection and Correction +/- 5 Degree

6. Character Recognition Text Characters varying from


14 points to 36 points

7. Additional Features: Reduces speckles and smudge


Recognizes different papers such as
bond, glossy,photocopier etc.
Supports Offset prints, laser prints
and
photocopy of offset and laser prints

Potrebbero piacerti anche