|
Functionality:
-
The status bar of the AVTOCRTrainfileBrowser now displays
information on the sample that is selected in the sheet.
-
In the AVTOCRTrainfileBrowser, some characters could not
be distinguished, e.g., the lower-case "l" and the
upper-case "I". The trainfile browser now attempts to
use the font "Times New Roman" for the symbol
tree and sheet, which allows to better distinguish
critical characters like lower-case "l" and upper-case
"I". The new AVTProperty "TreeSheetFontName" allows to
specify a different font name. However, a font is only
allowed if it supports all characters used by the
system locale, which is why "Times New Roman" will not be
used on Asian systems. In this case, the normal font
used for other GUI elements is used.
-
In the AVTOCRTrainfileBrowser, the symbol tree is now
alphabetically resorted immediately after a symbol is
renamed.
-
In the AVTOCRTrainfileBrowser, it was only possible to
copy a sample by drag and drop from the sheet to a new
trainfile. Now, it is also possible to move it.
-
If the property "ShowIconScaled" of the
AVTOCRTrainfileBrowser was "True", the display was scaled
so that the brightest gray value was white and the
darkest gray value was black. Now, in accordance with
the scaling which is actually used by the OCR algorithm,
the brightest 15% of all pixels are treated as
white, and the darkest 15% of all pixels as black.
-
Bug Fixes:
-
The segmentation of AVTOCR did not work with color
images. Now, it processes the red channel like the other
ActivVisionTools.
-
If an OCR font without trainfile was used, and if the *.ocr
file was physically write protected, this caused an
error although no data needed to be written. This
problem has been fixed.
-
If an OCR font with only one symbol was used, a "?"
instead of the symbol was read. Furthermore, the
behavior of the AVTOCRTrainfileBrowser was not correct.
These problems have been fixed.
-
Sometimes, the read symbols were not displayed in the
correct position under the respective objects. This
problem has been fixed.
-
If the segmentation of the OCR supplied no objects,
AVTOCRTeach continued to display the highlighted symbol
and taught text from the last segmentation result. This
problem has been fixed.
-
If the AVTOCRTrainfileBrowser was opened with only a
trainfile that does not belong to the selected OCR font, the
trainfile itself was selected instead of the first symbol of the
trainfile. This problem has been fixed.
-
In the AVTOCRTrainfileBrowser, the button "Remove" was
activ if a trainfile was selected that belongs to the
selected OCR font although those files cannot be
removed. Now, the button "Remove" is disabled if the
selected file belongs to the selected OCR font.
-
If, in the AVTOCRTrainfileBrowser, a sample was moved by
drag and drop within one trainfile from one symbol to
another symbol, the sheet still displayed the moved
sample. This problem has been fixed.
-
If, in the AVTOCRTrainfileBrowser, a sample was moved
from one trainfile to another by drag and drop or by
cutting and pasting it with CTRL-X and CTRL-V, the
sample was not correctly removed from the source
trainfile. This problem has been fixed.
- If, in the AVTOCRTrainfileBrowser, a sample was moved
or copied from a trainfile to the trainfile that belongs
to the selected OCR font the error message "Error
writing trainfile" appeared if the destination trainfile
was empty. This problem has been fixed.
-
If a new OCR font was selected while the
AVTOCRTrainfileBrowser was open, the columns "Read
Symbol" and "Next Best" were not updated correctly.
This problem has been fixed.
-
Moving the last sample of a trainfile to another symbol
could cause a crash. This problem has been fixed.
-
In the AVTOCRTrainfileBrowser, the data in the sheet was
not correctly updated when the name of a symbol was
changed in the tree.
-
If the "Save As ..." button was used to overwrite the current
font itself, the trainfile was lost in the process. This
problem has been fixed.
|
Release Notes for Version 3.0.1
ActivVisionTools Homepage
