Nuance has been upgraded to version 19.0 from 18.6.5.


Nuance v 19 change logs:

OmniPage Capture SDK Change Log

2014-11-22 Version 19.0 (14572.2000)


A major release!


2014-07-21 Version 18.6.7 (14371.1400)


Bug fixes:

– Crashes due to PNG files without DPI information fixed (bug #51342)

– Image specific buffer overrun fixed (bug #51342)

– Image specific stack overflow fixed (bug #53934)


– The barcode type Patch Code added

2014-05-12 Version 18.6.6 (14263.100)


Bug fixes:

– Image specific floating point exception and crash fixed (bug #51342)

– Fixed handling very small (few pixel) images (bug #52999)

– Image specific infinite loop and crash fixed (bug #53013)


– Handling zone files created with version 19 added

2013-11-11 Version 18.6.5 (13561.1637)


– Fixed a bug at zip file handling causing sigsegv mostly when used from Java

– Small fixes in Zoning, PDF font handling, FireWrx, and Mng subline handling

2013-01-22 oplicmgr


Fixed a bug in oplicmgr that removed the license file if append failed

2012-12-12 Version 18.6 (12612.1800)


A major release!

Most important addition is 64 bit support

2012-12-10 Version 18.6-Beta4 (12610.1316)


It’s a 32-bit release

– BAR and OMR modules supported in Linux, too

– COLOR_ARGB supported

– PDF glyph rendering warns if the correct font is not installed

2012-11-19 Version 18.6-Beta (12563.1227)


External beta of the 64-bit release of 18.6


2012-11-20 Version 18.5.3 (12570.100)


– Zone detection fixed when forced to return small zones

2012-11-06 Version 18.5.2 (12556.1115)


Bug fixes:

– Color space handling in JPEG compressed TIFF files fixed


– Zone order returned by kRecLocateZones improved

2012-09-25 Version 18.5.1 (12475.100)


– Small fixes in deskew and jbig2

2012-09-19 Version 18.5 (12469.100)


A major release! Most important additions:

– The Plus3W (and MOR) engines are supported

– The RecPDF module is added


2011-01-20 Version (11070.100)


Bug fixes:

– Black and White PNG file input inverted the image; fixed

– Some memory leaks when recognizing a particular Asian image fixed

– Segmentation fault caused by a particular image with a single Korean character fixed


– Some Asian accuracy improvements

2010-12-9 Version (10609.100)


Bug fixes:

– Handling of JBIG2 compressed PDF images fixed (Incident #56901)

– Some memory leaks fixed (Incident #60579)

– Removed a global symbol “filename” that collided with the libinifile library (Incident #57925)


– DTXT_PDFIOT output supports saving the original (not rotated) image when the

Kernel.Img.KeepOriginalImage setting is true. (Incident #58157)

– Small Korean recognition accuracy improvements

2010-8-17 Version 15.6.1 (10417.100)



– Improved detection of Simplified Chinese / Traditional Chinese / Japanese

languages when the input is a single line of Asian text. This functionality

can be enabled with the Kernel.OcrMgr.Asian.EnableCCJRecognition setting.

2010-7-9 Version 15.6.0 (10359.100)


Bug fixes:

– Segmentation fault during recognition of a particular, heavily broken, image

(Incident #53800)

– Image only PDF containing some empty lines of text is treated as textual and

generated no output (Incident #56901)

– Segmentation fault during recognition of different Japanese, English, French,

Spanish and German images.

– Arithmetic exception during recognition of different Japanese and English images.

– Page Description is not used with Asian languages

– Some gridded tables an Asian documents are not processed correctly

– PDF text with strangely named embedded font is not detected as italic


– JBIG2 compression added, JBIG2 compressed PDF images are handled now

– kRecSetLicense can load the license file from a relative path

– New API functions: kRecManageLanguages, kRecFindLanguageEx

– New functionality: kRecSetLanguageDetection, kRecGetPageLanguages

2009-7-13 Version 15.5.4 (9359.100)


Bug fixes:

– Race condition between two CSDK processes could result in deletion of all files

in the current working directory

– Segmentation fault during recognition of a specific Japanese image

– Segmentation fault during recognition of a specific English image

– Divide by zero error during recognition of a specific Japanese image

– API_GPFAULT_ERR returned during recognition of a specific Japanese image

– DTXT_XMLCOORD output containing an unpaired </dd> tag for an empty image

– DTXT_TXTF missing text (mostly following a bullet)

– DTXT_TXTF second half of text is garbage (many NULL characters)

– IMF_COMP_ERR due to a 260 byte limit in PDF file names (limit increased to 4096)


– Zones are re-ordered after recognition to follow basic text reading order rules.

Affects DTXT_TXTS output mostly.

– Asian output can be in UTF-8 format as well

– Automatic zoning of an empty image returns a warning, not an error.

Consequence: empty output file is created

– The Linux RPM packages are relocatable. Prefix = /usr/local

Following file contains Nuance v 19 change logs: