Legacy ImageMagick Discussions Archive

I can use IrfanView to convert a pdf into a multipage tiff file.
This file can be imported into OmniPage (then OCRed).

If I use ImageMagick to convert the same pdf into a multipage tiff file, then the file cannot be imported into OmniPage.
OmniPage complains - "OmniPage SE cannot load or scan this image.".

I use this command which results in a tiff with a similar filesize to the one produced by IrfanView.

convert -density 96x96 myPdfDoc.pdf -resize 1056x816 myTifDoc.tif

The file can be opened by other applications, eg: Microsoft Office Document Imaging, Microsoft Picture and Fax Viewer, IrfanView.
If I open it in IrfanView and then resave it, then OmniPage is happy with it.

What is going on?

ImageMagick creates a valid TIFF image but one that OmniPage cannot read. To fix, you need to identify what TIFF types OmniPage supports and add options to the ImageMagick command line to produce that TIFF type. Post the output of the

• identify -verbose

command for the image ImageMagick creates and the one Infraview creates. Perhaps we can figure out the requirements from the differences in image attributes.

Thanks for the prompt reply!
Here are the first image section snippets from the output of

• $identify -verbose pdf2TiffWithIM.tif
  
  Image: pdf2TiffWithIM.tif
  Format: TIFF (Tagged Image File Format)
  Class: DirectClass
  Geometry: 631x816+0+0
  Resolution: 96x96
  Print size: 6.57292x8.5
  Units: Undefined
  Type: TrueColor
  Base type: TrueColor
  Endianess: MSB
  Colorspace: RGB
  Depth: 16-bit
  Channel depth:
  red: 16-bit
  green: 16-bit
  blue: 16-bit
  Channel statistics:
  red:
  min: 0 (0)
  max: 65535 (1)
  mean: 44559.9 (0.67994)
  standard deviation: 18432.9 (0.281268)
  kurtosis: -1.12467
  skewness: -0.0634544
  green:
  min: 0 (0)
  max: 65535 (1)
  mean: 46875.9 (0.715281)
  standard deviation: 16544.7 (0.252457)
  kurtosis: -0.533182
  skewness: -0.234325
  blue:
  min: 0 (0)
  max: 65535 (1)
  mean: 61374.2 (0.93651)
  standard deviation: 12525.3 (0.191123)
  kurtosis: 14.7549
  skewness: -3.89405
  Image statistics:
  Overall:
  min: 0 (0)
  max: 65535 (1)
  mean: 38202.5 (0.582933)
  standard deviation: 26843.9 (0.409612)
  kurtosis: -1.45017
  skewness: -0.349018
  Rendering intent: Undefined
  Interlace: None
  Background color: white
  Border color: rgb(223,223,223)
  Matte color: grey74
  Transparent color: black
  Compose: Over
  Page geometry: 631x816+0+0
  Dispose: Undefined
  Iterations: 0
  Scene: 0 of 133
  Compression: None
  Orientation: TopLeft
  Properties:
  date:create: 2009-12-08T11:58:05+00:00
  date:modify: 2009-12-08T11:58:52+00:00
  signature: 1b9949721ad8d7975a8be9dd8c797ad0c0350bbb514d92fd6e82eb11d57a3075
  tiff:document: pdf2TiffWithIM.tif
  tiff:photometric: RGB
  tiff:rows-per-strip: 2
  tiff:software: ImageMagick 6.5.8-3 2009-11-28 Q16 http://www.imagemagick.org
  Artifacts:
  verbose: true
  Tainted: False
  Filesize: 365.8MiB
  Number pixels: 503KiB
  Pixels per second: 264KiB
  User time: 1.906u
  Elapsed time: 0:02.906
  Version: ImageMagick 6.5.8-3 2009-11-28 Q16 http://www.imagemagick.org

• $identify -verbose pdf2TiffWithIrfanView.tif
  
  Image: pdf2TiffWithIrfanView.tif
  Format: TIFF (Tagged Image File Format)
  Class: DirectClass
  Geometry: 816x1056+0+0
  Resolution: 96x96
  Print size: 8.5x11
  Units: PixelsPerInch
  Type: TrueColor
  Base type: TrueColor
  Endianess: MSB
  Colorspace: RGB
  Depth: 8-bit
  Channel depth:
  red: 8-bit
  green: 8-bit
  blue: 8-bit
  Channel statistics:
  red:
  min: 0 (0)
  max: 255 (1)
  mean: 173.426 (0.6801)
  standard deviation: 73.0236 (0.286367)
  kurtosis: -1.03044
  skewness: -0.125362
  green:
  min: 0 (0)
  max: 255 (1)
  mean: 182.436 (0.715437)
  standard deviation: 65.7418 (0.257811)
  kurtosis: -0.354149
  skewness: -0.321318
  blue:
  min: 0 (0)
  max: 255 (1)
  mean: 238.832 (0.936594)
  standard deviation: 50.467 (0.19791)
  kurtosis: 15.0705
  skewness: -3.96344
  Image statistics:
  Overall:
  min: 0 (0)
  max: 255 (1)
  mean: 148.673 (0.583033)
  standard deviation: 105.103 (0.412167)
  kurtosis: -1.46045
  skewness: -0.350269
  Rendering intent: Undefined
  Interlace: None
  Background color: white
  Border color: rgb(223,223,223)
  Matte color: grey74
  Transparent color: black
  Compose: Over
  Page geometry: 816x1056+0+0
  Dispose: Undefined
  Iterations: 0
  Scene: 0 of 133
  Compression: None
  Orientation: TopLeft
  Properties:
  date:create: 2009-12-08T11:48:14+00:00
  date:modify: 2009-12-08T11:49:43+00:00
  signature: 0748cf4c66b909da4b45afe14169d71560d81a5d8ec90e0182970010c220881c
  tiff:photometric: RGB
  tiff:rows-per-strip: 3
  tiff:software: IrfanView
  Artifacts:
  verbose: true
  Tainted: False
  Filesize: 328.3MiB
  Number pixels: 842KiB
  Pixels per second: 15.7KiB
  User time: 5.438u
  Elapsed time: 0:54.631
  Version: ImageMagick 6.5.8-3 2009-11-28 Q16 http://www.imagemagick.org

The output from the verbose identify command on the IrfanView generated tif is 2344KB whereas the output from the ImageMagick generated tif is only 258KB.
Do you want me to upload both files so you can 'diff' them?

ImageMagick produces 16-bit per channel TIFF images. Let's start with that. Add

• -depth 8

to your command line.

Yes, I saw that and have already tried it.
It does make it work but the resolution is awful! Worse than with IrfanView.
I thought I'd try ImageMagick to see if I could get better resolution than with IrfanView.

I use OmniPage for OCRing so I'd like optimum resolution (ideally 300x300 according to the OmniPage manual).
I don't really understand how to achieve that from the original pdf.

I did an identify on the pdf and it says -

Image: myPDFDoc.pdf
Format: PDF (Portable Document Format)
Class: DirectClass
Geometry: 612x792+0+0
Resolution: 72x72
Print size: 8.5x11
Units: Undefined
Type: TrueColor
Endianess: Undefined
Colorspace: RGB
Depth: 16/8-bit
Channel depth:
red: 8-bit
green: 8-bit
blue: 8-bit
Channel statistics:
red:
min: 0 (0)
max: 65535 (1)
mean: 44555 (0.679866)
standard deviation: 18776.5 (0.286511)
kurtosis: -1.02682
skewness: -0.126939
green:
min: 0 (0)
max: 65535 (1)
mean: 46879.1 (0.71533)
standard deviation: 16890.2 (0.257728)
kurtosis: -0.352392
skewness: -0.321098
blue:
min: 0 (0)
max: 65535 (1)
mean: 61340.1 (0.93599)
standard deviation: 13007.6 (0.198484)
kurtosis: 14.8609
skewness: -3.93556
Image statistics:
Overall:
min: 0 (0)
max: 65535 (1)
mean: 38193.6 (0.582796)
standard deviation: 27006.7 (0.412096)
kurtosis: -1.46048
skewness: -0.349525
Rendering intent: Undefined
Interlace: None
Background color: white
Border color: rgb(223,223,223)
Matte color: grey74
Transparent color: black
Compose: Over
Page geometry: 612x792+0+0
Dispose: Undefined
Iterations: 0
Scene: 0 of 133
Compression: Undefined
Orientation: Undefined
Properties:
date:create: 2009-12-08T14:27:20+00:00
date:modify: 2009-12-08T14:28:19+00:00
pdf:HiResBoundingBox: 612x792+0+0
pdf:Version: PDF-1.4
signature: 53b63bf32cc565833b88617295723381714ffd823a3c95a179de302998c6b6cc
Artifacts:
verbose: true
Tainted: False
Filesize: 183.1MiB
Number pixels: 473KiB
Pixels per second: 522KiB
User time: 0.891u
Elapsed time: 0:01.905
Version: ImageMagick 6.5.8-3 2009-11-28 Q16 http://www.imagemagick.org

don't know if this will help, but your IM print size is different from your Irfan and your IM units are undefined. try adding -units PixelsPerInch after -density

see http://www.imagemagick.org/script/comma ... .php#units

I'm just trying gsView to achieve the same goal.
I can open the pdf and convert using a device such as tiff24nc (24-bit RGB, 8-bits per component) and select the max available resolution of 204x196.
When the resultant Tiff is opened in OmniPage SE it says

• Image Resolution is 98,102 (originally 196, 204)
• Image Size is 1146x843

The Tiff's filesize is rather large (over 1GB!).
I then use OmniPage SE to save each page as a jpeg (approx 100-200 KB each).
Canon MP Navigator 2.0 then converts all these jpegs into a pdf of 16MB.
Unfortunately it doesn't seem to be searchable! Arghh! So close!

I've had some success with ImageMagick now.

My latest command is :-

convert -depth 8 -density 96x96 -units PixelsPerInch file.pdf -resize 816x1056 myTifDoc.tif

I swapped from 1056x816 to 816x1056 and the OCR recognition suddenly began working again!

I've now tried IrfanView, GSview and ImageMagick and so far the later with the command above seems to give the best OCR results.

Thanks everyone.

If anyone can suggest how I might improve the OCR recognition further please shout!
I tried
convert -depth 8 -density 150x150 -units PixelsPerInch file.pdf -resize 816x1056 myTifDoc.tif
but it seemed to be slightly worse than with the lower density of 96x96.