[ocrfeeder/user_documentation] Configuration added
- From: Joaquim Manuel Pereira Rocha <jrocha src gnome org>
- To: commits-list gnome org
- Cc:
- Subject: [ocrfeeder/user_documentation] Configuration added
- Date: Mon, 13 Dec 2010 15:22:21 +0000 (UTC)
commit b3cd9953f16cfc79d7224192798d6542abb594fb
Author: Joaquim Rocha <jrocha igalia com>
Date: Mon Dec 13 16:22:05 2010 +0100
Configuration added
help/C/deskewing.page | 29 ++++++++++++++
help/C/finetuning.page | 54 ++++++++++++++++++++++++++
help/C/index.page | 4 ++
help/C/ocrconfiguration.page | 87 ++++++++++++++++++++++++++++++++++++++++++
help/C/unpaper.page | 45 +++++++++++++++++++++
5 files changed, 219 insertions(+), 0 deletions(-)
---
diff --git a/help/C/deskewing.page b/help/C/deskewing.page
new file mode 100644
index 0000000..00fbe39
--- /dev/null
+++ b/help/C/deskewing.page
@@ -0,0 +1,29 @@
+<page xmlns="http://projectmallard.org/1.0/"
+ type="topic"
+ id="deskewing">
+
+<info>
+ <link type="guide" xref="index#configuration"/>
+ <link type="seealso" xref="manualeditionandcorrection"/>
+ <desc>Correcting the skew in the images</desc>
+</info>
+
+<title>Deskewing</title>
+
+<p>Some images, especially if they were added from a scanner device,
+may be skewed and this makes it harder to recognize the image.</p>
+
+<p><app>OCRFeeder</app> offers a way to automatically deskew an
+image. To deskew a loaded image, click
+<guiseq><gui>Tools</gui><gui>Image Deskewer</gui></guiseq>.</p>
+
+<p>This operation can also be set to be performed automatically
+every time an image is added. To set it, simply open the
+<gui>Preferences</gui> dialog from
+<guiseq><gui>Edit</gui><gui>Preferences</gui></guiseq> and check
+<gui>Deskew images</gui> under the <gui>Tools</gui> tab.</p>
+
+<note type="warning"><p>Depending on the size and characteristics
+of the image, deskewing an image may take some time.</p></note>
+
+</page>
diff --git a/help/C/finetuning.page b/help/C/finetuning.page
new file mode 100644
index 0000000..1a3d143
--- /dev/null
+++ b/help/C/finetuning.page
@@ -0,0 +1,54 @@
+<page xmlns="http://projectmallard.org/1.0/"
+ type="topic"
+ id="finetuning">
+
+<info>
+ <link type="guide" xref="index#configuration"/>
+ <link type="seealso" xref="manualeditionandcorrection"/>
+ <desc>Advanced options for a better recognition</desc>
+</info>
+
+<title>Fine-tuning</title>
+
+<p><app>OCRFeeder</app> has some advanced options that can be
+used to perform a better recognition. These options can be
+chosen from the <guiseq><gui>Edit</gui><gui>Preferences</gui></guiseq>
+dialog, under its <gui>Recognition</gui> tab.</p>
+
+<p>The following list describes the mentioned options:</p>
+<list>
+ <item><p><gui>Fix line breaks and hyphenization</gui>: OCR engines
+ usually read the text line by line and seperate each line with a
+ line break. Sometimes, this is not what the user wants because the
+ text might be broken in the middle of a sentence.</p>
+ <p>Checking this option will make <app>OCRFeeder</app> remove single
+ newline characters after the text is recognized by the engines.</p>
+ <p>Since just removing newlines in an hyphenized text would result
+ in wrongly separated words, hyphenization is also detected and removed
+ in this process.</p></item>
+ <item><p><gui>Window Size</gui>: <app>OCRFeeder</app>'s algorithm to
+ detect the contents in an image uses the concept of <em>window size</em>
+ which is the division of the image in small windows. A smaller window
+ size means it is likely to detect more content areas but size that is
+ too small may result in contents that should be part of a bigger area
+ instead. On the other hand, a bigger window size means less divisions
+ of contents but may end up in contents which should be subdivided.</p>
+ <p>A good window size should be slightly bigger than the text line spacing
+ in an image.</p><p>Users may want to manually set this value if automatic
+ one doesn't produce any valid content areas but normally it is easier to
+ use the automatic one and perform any needed corrections directly in
+ the content areas.</p></item>
+ <item><p><gui>Improve columns detection</gui>: Check this option if
+ <app>OCRFeeder</app> should try to divide the detected content areas
+ horizontally (originating more columns). The value that is used to
+ check the existance of blank space within the contents may be set to
+ automatic or manual when the columns aren't detected correctly.</p></item>
+ <item><p><gui>Adjust content areas' bounds</gui>: The detected content
+ areas sometimes have a considerable margin between their contents and
+ the areas' edges. By checking this option, <app>OCRFeeder</app> will
+ minimize those margins, adjusting the areas to its contents better.
+ Optionally, a manual value can be check to indicate the minimum value
+ of the adjusted margins.</p></item>
+</list>
+
+</page>
diff --git a/help/C/index.page b/help/C/index.page
index 8f02a7f..648b78c 100644
--- a/help/C/index.page
+++ b/help/C/index.page
@@ -35,4 +35,8 @@ its main one ODT.</p>
<title>Recognition</title>
</section>
+<section id="configuration" style="2column">
+ <title>Configuration</title>
+</section>
+
</page>
diff --git a/help/C/ocrconfiguration.page b/help/C/ocrconfiguration.page
new file mode 100644
index 0000000..2cd059a
--- /dev/null
+++ b/help/C/ocrconfiguration.page
@@ -0,0 +1,87 @@
+<page xmlns="http://projectmallard.org/1.0/"
+ type="topic"
+ id="ocrconfigutation">
+
+<info>
+ <link type="guide" xref="index#configuration"/>
+ <link type="seealso" xref="automaticrecognition"/>
+ <link type="seealso" xref="manualeditionandcorrection"/>
+ <desc>Configure the OCR engines to recognize the text</desc>
+</info>
+
+<title>OCR Engines Configuration</title>
+
+<p><app>OCRFeeder</app> uses system-wide OCR engines to extract
+the text from images. This means any OCR engine that can be
+used from the command line should also be used in <app>OCRFeeder</app>.</p>
+
+<section>
+
+<title>Automatic Detection of OCR Engines</title>
+
+<p>The OCR engines (<em>Tesseract</em>, <em>GOCR</em>, <em>Ocrad</em>
+and <em>Cuneiform</em>) are already automatically detected and configured
+in most systems, the first time <app>OCRFeeder</app> is run.</p>
+
+<p>If an OCR engine is installed after <app>OCRFeeder</app> had configured
+already an engine, it will not be automatically configured but, depending on
+the engine, users might easily go to the <gui>OCR Engines</gui> dialog and
+choose it from the list of detected engines after pressing <gui>Detect</gui>.</p>
+
+<note style="tip"><p>Already configured OCR engines might be detected again and it is
+up to the user to uncheck these engines if they shouldn't be added again.</p></note>
+
+</section>
+
+<section>
+
+<title>Manual Configuration</title>
+
+<p>The currently configured OCR engines are shown in the
+<gui>OCR Engines</gui> dialog which can be opened from
+<guiseq><gui>Tools</gui><gui>OCR Engines</gui></guiseq>.</p>
+
+<p>Besides seeing the configured OCR engines, the <gui>OCR Engines</gui>
+dialog allows to add new engines, edit or delete the current ones and
+detect engines installed in the system.</p>
+
+<p>When adding or editing an OCR engine (by pressing the <gui>Add</gui>
+or <gui>Edit</gui> buttons, respectively), a dialog is shown with the
+following fields:</p>
+
+<list>
+ <item><p><gui>Name</gui>: The engine's name. This name will be used
+ in throughout the UI when referring to the engine;</p></item>
+ <item><p><gui>Image format</gui>: The image format that the engine
+ recognizes (for example, <em>TIF</em> in the case of
+ <em>Tesseract</em>);</p></item>
+ <item><p><gui>Failure string</gui>: Some engines replace unrecognized
+ characters by another, pre-defined character (for example,
+ <em>_</em> in the case of <em>GOCR</em>).</p></item>
+ <item><p><gui>Engine path</gui>: The path in the system to the
+ engine's executable (for example, <em>/usr/bin/tesseract</em>).</p></item>
+ <item><p><gui>Engine arguments</gui>: The arguments that feed an image
+ to the engine and make it output the recognized text to the standard
+ output. <app>OCRFeeder</app> runs the engine with these arguments
+ as if it was in the command line and looks for the recognized text
+ in the standard output. Some engines already do this, like
+ <em>Ocrad</em> and <em>GOCR</em> while other, like <em>Tesseract</em>,
+ write the text into a file.</p>
+ <p>Since the image's path to be read is always needed, a special argument
+ <em>$IMAGE</em> is provided for this and will be replaced by the image path
+ when the engine is run. For the cases
+ where a file name is needed, like the one mentioned previously, a special
+ argument <em>$FILE</em> is provided and will be replaced by a temporary
+ file name.</p>
+ <p>So, in case of <em>Tesseract</em> (which writes the recognized text
+ into a file), the arguments would be <em>$IMAGE $FILE; cat $FILE.txt;
+ rm $FILE</em>.</p></item>
+
+</list>
+
+<note style="advanced"><p>The engines' configuration is stored in their own XML file
+in the user's home under <em>.ocrfeeder/engines/</em>.</p></note>
+
+</section>
+
+</page>
diff --git a/help/C/unpaper.page b/help/C/unpaper.page
new file mode 100644
index 0000000..96189ce
--- /dev/null
+++ b/help/C/unpaper.page
@@ -0,0 +1,45 @@
+<page xmlns="http://projectmallard.org/1.0/"
+ type="topic"
+ id="unpaper">
+
+<info>
+ <link type="guide" xref="index#configuration"/>
+ <link type="seealso" xref="manualeditionandcorrection"/>
+ <desc>Cleaning images before performing OCR</desc>
+</info>
+
+<title>Unpaper</title>
+
+<p><em>Unpaper</em> is a tool to clean images in order to make
+them easier to read on screen. It is aimed mainly at images
+obtained from scanned documents which usually show dust, black
+margins or other flaws.</p>
+
+<p><app>OCRFeeder</app> can use <em>Unpaper</em> to clean its
+images before processing them, which usually results in a better
+recognition.</p>
+
+<p><em>Unpaper</em> needs to be installed in order to be used.
+If it is not installed, <app>OCRFeeder</app> won't show it's action
+in the interface.</p>
+
+<p>To use <em>Unpaper</em> on a loaded image, click
+<guiseq><gui>Tools</gui><gui>Unpaper</gui></guiseq>. The
+<gui>Unpaper Image Processor</gui> dialog will be shown with
+<em>Unpaper</em>'s options and an area to preview the changes before
+applying them to the loaded image. Depending on the size and
+characteristics of the image, using this tool might take some time.</p>
+
+<p><em>Unpaper</em> can be configured opening
+<guiseq><gui>Edit</gui><gui>Preferences</gui></guiseq> and accessing
+the <gui>Tools</gui> tab. In this area one can enter the path to
+<em>Unpaper</em>'s executable (normally this is already configured if
+<em>Unpaper</em> was installed the first time <app>OCRFeeder</app> was
+run). In the same area, under <gui>Image Pre-Processing</gui>, one can
+check <gui>Unpaper images</gui> to make images being processed automatically
+by <em>Unpaper</em> after they are loaded into <app>OCRFeeder</app>.
+The options taken by <em>Unpaper</em> when it's automatically
+called after adding an image can be configured by clicking the
+<gui>Unpaper Preferences</gui> button.</p>
+
+</page>
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]