Home > Tactile Graphics Assistant
Tactile Graphics
Assistant Software
Reference
& Usage Manual
Prepared By:
The Tactile Graphics Project
University of Washington
July 2007
Revised April, 2013
Table
of Contents
About the TGA 3
Tactile Graphics Workflow 3
Reference
Overviews
TGA Main Screen Overview 4
File Menu Overview 5
General Options Dialog Overview 6
Mode Menu Overview 7
Train Menu Overview 8
Execute Menu Overview 10
View Menu
Overview 11
Usage Tips
Selecting Characters12
Selecting Labels 13
Step-by-Step
Usage Guides
How to Use the TGA 14
Using the TGA as Part of the Image Translation Process 18
Preprocessing Images with
Adobe Photoshop 26
Appendices
About the Tactile Graphics Project 30
Modifications to Use QR-Codes 30
Alternative Software 31
Available Books 31
Contact Information 32
About the TGA
The Tactile
Graphics Assistant (TGA) is a software tool created by members of the
Tactile Graphics Project to assist in the text removal stage of the
visual-to-tactile image conversion process. Currently, the primary
function of the TGA is to detect and remove text from large batches
of digital images. The TGA takes advantage of the fact that images
from the same textbook possess many similar qualities, such as text
style, shading, and label placement. A small subset of images
is selected from the textbook to form a training set. The user
indicates the location of the text – individual characters as well
as meaningful labels – in each image. The TGA will then be ��trained��
on this data. The TGA then uses the training set data to automatically
detect characters and labels in the remaining images. If errors
occur during this procedure, the user can make changes to the TGA��s
selections. Once text removal has been successfully completed,
the TGA will output files that can be used in other stages of the image
transcription process.
Tactile Graphics Workflow
In order to create tactile graphics using the TGA, you will need several other pieces of software:
It is important to note that
currently the TGA workflow has only been tested with the software mentioned
above. If you use other image editing software, for instance, the scripts
designed for Photoshop and Illustrator will not work.
TGA Main Screen Overview
The main screen
of the TGA is where images are displayed and processed. This is
what the main screen looks like when the TGA is first opened:
The main screen
of the TGA.
When an image
is loaded, it appears in the display area of the main screen (see image
below). If the image does not fit in the display area, scroll
bars will appear. The name of the image file appears in the heading
bar at the top of the screen.
Notice in the
screenshot below that the ��Character Mode�� radio button is selected.
When images are opened, the TGA automatically switches to Character
Mode.
The TGA main screen with an image loaded.
File Menu
Overview
This section provides an overview
of all the commands listed in the File menu (opened below).
The File
menu.
General
Options: See detailed description on next page.
Allows the user to configure TGA settings including paper size, DPI,
and image folders.
Load:
Enables the user to choose an image to load into the TGA. A dialog
box will appear with browsing options.
Load
Previous: Loads the previous image in the designated input
folder (set in General Options).
Load
Next: Loads the next image in the designated input folder
(set in General Options).
Save:
Saves character and label selections in the current image. Saved
data is stored in the designated intermediate folder (set in General
Options).
Close:
Closes the current image.
Batch
Process: Automatically identifies characters and labels
in all images in the designated input folder (set in General
Options), generates output for each image, and then saves the changes
made. This is the equivalent of manually opening every image in
the folder and clicking the Mark Characters & Labels command in
the Execute menu. Currently it is not possible to cancel a batch process.
Exit: Exits the TGA. All training data accumulated in this session will be lost. Individual image data will be retained.
General
Options Dialog Overview
The General
Options dialog box can be accessed from the General Options
command in the File menu.
The General
Options dialog box.
Page
width in inches: Specifies the paper width.
Page
height in inches: Specifies the paper height.
Target
dots per inch (DPI): Specifies the DPI.
Preserve
Aspect Ratio: Specifies whether or not to preserve the
width/height ratio of the image on resize.
Input:
Enables the user to designate a specific folder as the ��input�� folder.
When the Batch Process command is selected, the TGA will automatically
open and process each image in the designated ��input�� folder.
In addition, when Load is selected, the browsing dialog box will
display the designated ��input�� folder.
Intermed:
Enables the user to designate a specific folder as the ��intermediate��
folder. When the Save command is selected, the data for
the current image will be saved in this folder.
Output:
Enables the user to designate a specific folder as the ��output��
folder. When the Mark Similar
or Mark Characters & Labels
commands are selected, the output files generated for each image will
be saved in this folder.
Levels of undo: Specifies the number of previous program states stored in the undo database. If ��3�� is entered, the user can select the Undo command up to three times, enabling him or her to undo the past three actions.
Mode Menu
Overview
This section
provides an overview of all the commands listed in the Mode menu
(opened below).
The Mode
menu.
As long as
an image is opened in the image display window, the TGA will be in either
Character Mode or Label Mode. The radio buttons on the main screen
perform the same functions as the commands in the Mode menu (see
image below).
Character Selection: Enables the user to designate elements in the current image as characters.
Label Selection:
Enables the user to group characters into text labels.
The mode selector radio buttons enable the user to switch between Character Mode and Label Mode.
Train Menu
Overview
This section
provides an overview of all the commands listed in the Train
menu (opened in the image below).
The Train
menu.
Update
Positive: Available in both Character Mode and Label Mode.
Updates the training database with positive examples selected.
If this command is selected in Character Mode, everything in the image
that has been selected as a character will be added to the training
database as examples of graphical elements that are characters.
If this command is selected in Label Mode, groupings of characters that
have been selected as labels will be added to the training database
as examples of labels.
Update
Negative: Only available in Character Mode. Updates
the training database with negative character examples. If this
command is selected, everything in the image that has not been
selected as a character will be added to the training database as examples
of graphical elements that are not characters.
Update
Positive and Negative: Only available in Character Mode.
Updates the training database with both positive and negative character
examples. If this command is selected, everything in the image
that has been selected as a character will be added to the training
data base as examples of graphical elements that are characters,
and everything that has not been selected as a character will
be added to the training database as examples of graphical elements
that are not characters.
Update All: Characters & Labels: Available in both Character Mode and Label Mode. Updates the training data base with both positive and negative character examples and examples of labels. If this command is selected, everything in the image that has been selected as a character will be added to the training data base as examples of graphical elements that are characters, and everything that has not been selected as a character will be added to the training database as examples of graphical elements that are not characters. Groupings of characters that have been selected as labels will be added to the training set as examples of labels.
Train
Menu Overview (continued)
Clear
Statistics: Clears either all characters or all
labels from the training set, depending on the current mode. Positive
and negative statistics sets are cleared.
Clear
All Statistics: Clears both characters and labels
from the training set. Positive and negative statistics sets are
cleared.
Options:
See image below. Enables the user to choose the color(s) of characters
identified by the TGA. The ��color picker�� box (on the left
side of the dialog) lists the color(s) of characters the user wishes
the TGA to identify. In the example below, ��0 0 0�� (representing
the color black) is the only color listed. To add colors to the
color picker list, click on a section of the image that contains the
desired color. To delete a color from the list, click on that
entry in the list and press the ��Delete Selected�� button.
The controls on the right side of the dialog box (labeled ��SVM/Old
Character Classification�� and ��Old Character Classification Only��)
were used for testing in the creation of the TGA and should not be modified.
The ��color picker�� dialog box.
Execute Menu
Overview
This section
provides an overview of all the commands listed in the Execute
menu (shown opened in the image below).
The Execute
menu.
Mark Similar:
Available in both Character Mode and Label Mode. If in Character
Mode, automatically identifies characters based on the current training
set. If in Label Mode, automatically identifies labels based on
the current training set.
Mark Characters &
Labels: Automatically identifies all characters and labels
based on the current training set.
Clear Marked:
Available in both Character Mode and Label Mode. If in Character
Mode, ��un-marks�� all selected characters. If in Label Mode,
��un-marks�� all selected labels.
Clear Characters &
Labels: Automatically ��un-marks�� all selected characters
and labels.
Undo: Erases the results of the last action completed.
View Menu
Overview
This section
provides an overview of all the commands listed in the View menu
(shown opened in the image below).
The View
menu.
State
Information: Opens a dialog box that lists information
about the current training set, characters and labels selected in the
current image, and other statistics.
Hide Selected Characters: When this option is checked, all the selected characters in the image are hidden. See image above. This option is also available directly from the main screen, by checking or unchecking the ��Hide Selected Characters�� checkbox.
Selecting
Characters
Select any
letters, numbers, and symbols in the image that you want to remove from
the image and translate to Braille. Once selected, these characters
will be enclosed in a red box on the main screen of the TGA. Don��t
forget to select punctuation, as well as the dots on ��i��s and ��j��s.
A red box indicates
letters/numbers/symbols designated as characters.
Selecting characters
is easy and can be done quickly. For large patches of text with
no interfering graphical elements, simply drag the mouse over the whole
patch (see image below); all the text will be selected as characters.
Large groups of letters and symbols can be selected as characters at the same time.
Be careful not to select any unwanted graphical elements.
Selecting
Labels
A set of characters
that will function as one Braille label should be grouped together as
a label. Once selected, labels will be enclosed in a blue box
on the main screen of the TGA. The characters in each label will
be output by the TGA together, so they can be directly sent to translation
software.
A blue box indicates
characters grouped as a label.
To group characters
into a label, drag the mouse over the group of selected characters you
want to designate as a label (see image below). The TGA will only
look for selected characters in the area you highlight, so it��s
OK if you accidentally highlight other non-character graphical elements
(as in the image below).
Complete accuracy is not required for label selection.
Only selected characters in the highlighted area will be grouped in the label.
In this image, the letters will be grouped as a label, but other graphical elements will be ignored.
How to Use the TGA
This section guides the user
through the process of editing a batch of images using the TGA.
Step 1: Set Up Images and Directories
Step 2: Configure TGA Settings
The ��General Options�� dialog box.
How to Use the TGA (continued)
Step 3: Load Images
Step 4: Character Selection
The ��color picker�� dialog box.
How to Use the TGA (continued)
Selecting characters
in an image.
Step 5: Label Selection
Grouping selected characters into a label.
How to Use the TGA (continued)
Step 6: Saving Images
Step 7: Training
Step 8: Batch Processing
Using the TGA as Part of
the Image Translation Process
The TGA was designed for use
as one component a tactile graphics translation workflow. In this
workflow, different software tools automate steps in the tactile graphics
translation process.
Using the TGA as Part of
the Image Translation Process
Step 1: Preliminary
Step 2: Classification
Step 3:
TGA – Repeat the steps for each class
(See the How to Use the
TGA section on page 14 for more details on each step)
Using the TGA as Part of
the Image Translation Process (continued)
Step 4:
OCR – Repeat the steps for each set
(We use OmniPage Pro OCR
software for this step. However, other OCR software can be used.
Most popular OCR software tools have similar batch processing capabilities.)
OmniPage OCR Batch Process
Step 1: Load Image Files
OmniPage OCR Batch Process Step 2: Recognize Images
Using the TGA as Part of
the Image Translation Process (continued)
Step 4:
OCR (continued)
OmniPage OCR Batch Process
Step 3: Save as OPD
OmniPage OCR Batch Process Step 4: Finish Job
Using the TGA as Part of
the Image Translation Process (continued)
Step 4:
OCR (continued)
How to find the ��Save to File�� command.
Using the TGA as Part of
the Image Translation Process (continued)
Step 4:
OCR (continued)
How to set up the ��Save to
File�� dialog box
Using the TGA as Part of
the Image Translation Process (continued)
Step 5: Braille Translation
– Repeat the steps for each set
(We use Duxbury Braille
Translator for this step. Other Braille translation software can
be used, but care must be taken to ensure that the resulting Braille
document has the same number of lines as the original.)
Step 6:
Resize Images – Repeat steps for each set
How to navigate to the ��script selection�� menu item.
Using the TGA as Part of
the Image Translation Process (continued)
Step 7:
Label Placement – Repeat steps for each set
Step 8:
Editing in Photoshop – Repeat for each set
Step 9:
Editing in Illustrator – Repeat for each set
Preprocessing Images With
Adobe Photoshop
Photoshop can be used to automatically
preprocess all images in a folder the same way. This step demonstrates
how to do this ��batch processing�� in Photoshop. Images can
also be preprocessed one at a time.
The ��New Action�� button (shown circled in the lower right) and the ��New Action�� popup window.
Preprocessing Images With
Adobe Photoshop (continued)
How to navigate to the ��Threshold��
dialog box.
The ��Threshold�� dialog box.
Preprocessing Images With
Adobe Photoshop (continued)
The ��Stop�� button is circled
in the ��Action�� window.
Preprocessing Images With
Adobe Photoshop (continued)
How to navigate to the ��Batch��
command.
The ��Batch�� dialog box.
Appendices
Project Overview
The Tactile
Graphics Project aims to increase universal benefit from graphical images
(i.e., line graphs, bar charts, illustrations, etc.). It is a multidisciplinary
project with researchers and practitioners from UW's Department of Computer
Science and Engineering, Access Technology Lab, and DO-IT. Our goal
is to enable K-12, college, undergrad, and graduates students who are
blind to have full access to mathematics, engineering, and science.
We have developed the Tactile Graphics Assistant software which, when
combined with off-the-shelf software applications, enables the rapid
translation of visual graphics to a tactile form.
Research includes:
Modifications to Use
QR Codes
An alternative to Braille for
text in images is to use QR-codes which are public code for which readers
are freely available. An example of a QR-code is in the figure below.
��This is an
example of a QR-code��
There are several advantages
of QR-codes over Braille. First, QR-codes generally take less
area than Braille, allowing for the inclusion of more text in a tactile
graphic. Second, QR-codes would allow these who do not know Braille
to have access to text in tactile graphics. A disadvantage of
QR-codes is the time to find and recover the text stored in them with
a smartphone QR-code reader. An example of a tactile graphic with QR-codes
instead of Braille is shown below.
A tactile graphic
with QR-codes.
In order to modify TGA to use
QR-codes instead of Braille, two steps in the TGA process need to be
replaced. Step 5 translates the text labels into Braille. This step
will need to be replaced with generating QR-codes for the text. Step
7 will need to be replaced with the QR-code placement. Currently the
QR-codes are manually being cut out and glued on top of the tactile
graphics in their appropriate locations. In the future it is possible
to print the QR codes using an ink printer directly. If this is done,
tactile markers need to be added to the image to allow the QR codes
to be found. A potential marker is ^ with the QR code lining up below
the opening.
Alternative Software
The software other that we
have used were the best available at the time we developed the Tactile
Graphics Workflow. The Adobe Creative Suite (Photoshop, Illustrator),
Omnipage, and Braille translation software are relatively expensive.
The make the workflow possible, the software should have scripting ability
so that many images can be processed in one run. New scripts would
have to be written to use any alternative software. Modifications
to the workflow may be necessary to accommodate these alternatives.
Below is a list of alternative software that is free and downloadable
from the web.
Available Books
Please visit our website for more information:
http://tactilegraphics.cs.washington.edu
Contact Information
The Tactile Graphics Project website:
http://tactilegraphics.cs.washington.edu
To download the executable for the TGA, please register at this website:
http://depts.washington.edu/ventures/UW_technology/Express_Licenses/tga.php
To obtain the source code for the TGA, please contact:
UW Tech Transfer Digital Ventures
UW Tech Transfer
4311 11th Avenue NE Suite 500
Seattle, WA 98105-3451
Phone: (206) 616-3451
Fax: (206) 616-3322
E-mail: license@u.washington.edu
All Rights Reserved Powered by Free Document Search and Download
Copyright © 2011