The command "X11Handwriting" creates a handwriting recognition widget.
The
Handwriting Recognition widget has four rows
1) A label displaying the recognized text and accepting gestures
2) A box row for drawing characters
3) A 1st command row
4) A 2nd command row for interoperation with intelligent tools
The label displaying the recognized text
Listens to some simple gestures:
-Strike out some characters (horizontal line)
-Set the "input focus" (tap)
-Stuff the X selection (checkmark)
-Change capitalization (/ pointing upwards)
-Make control character (\ pointing upwards)
The 1st command row for local command
Menu...
Shows the main pop up menu.
cap
Changes capitalization of last recognized character.
2nd
Use second choice of last recognized character.
Xset
Simulates keyboard entry of the prepared text with artifical events
(Also puts the text in the primary selection and the clipboard)
Xcr
Simulates a <Return> keyboard entry with artifical events
Xbs
Simulates a <Backspace> keyboard entry with artifical events
The 2nd command row for interoperation with intelligent tools
SET
Takes the recognized text and stuffs it into the primary selection and the clipboard
SET+
Like set but also adds a linefeed
r-clear
Clears the widget with the input focus
r-back
Backspace one character on the widget with the input focus.
enter
Performs "ENTER" operation on the widget with the input focus.
next
Like using the "NEXT" key on the widget with the input focus.
The main pop up menu
training
Opens a "handwriting recognition training" widget.
uni-strokes
Recognizes unistrokes characters
Side effect: switching to uni selects the default unistrokes training database
latin-alphabet
Recognizes multi stroke characters (e.g. the latin alphabet)
Side effect: switching to latin selects the default latin training database
pseudo keyboard
Creates a widget with a pseudo keyboard for the alphabetic characters
Different mouse or pen button for capitalization
extra keyboard
Creates a widget with a pseudo keyboard for the numeric and special characters
The
Handwriting Recognition
training widget has five rows
A REPLAY row for regressions and data aquisition
REPLAY
replays last entry from the regression database and remove it from the regression database.
remember
remembers the last strokes in the regression database
write
makes up a filename and write the regression database onto it
new
nilout the regression database
read
read the regression database from named file.
file
file name for loading the regression database
A DB row accessing the "training database"
DB: inert label
write: Make up a filename and write the training data base.
new: Nilout the training database. This needs to re-read of course.
merge: Read and merge in a training data base from named file.
file: File name used for reading the training database
A TRAIN row "training commands" row
TRAIN: Add the last strokes to the data base
forget1: Remove the last best "recognized" strokes from the data base
forget2: Remove the last second best "recognized" strokes from the data base
char: A character field for the character to be trained. This field has
an inoperative default to protect against the frequent error of training
an unspecified character.
3 character fields, "badness" and "character" values
The smaller a badness value, the better match was found.
1) The normalized strokes of the last recognized character
2) The strokes of the best match from last recognition
3) The strokes of the second best match from last recognition
The special X11Handwriting-TrainingMenu command
Use
X11Handwriting-TrainingMenu EntryName {FullPathFileName|Default}
To enter or remove options to the training database menu. If multiple filenames are used, databases will be merged.