do_ocr_single_class_cnn🔗
Short description🔗
do_ocr_single_class_cnn — Classify a single character with an CNN-based OCR classifier.
Signature🔗
do_ocr_single_class_cnn( region Character, image Image, ocr_cnn OCRHandle, integer Num, out string Class, out real Confidence )
Description🔗
do_ocr_single_class_cnn computes the best classification for the
character given by the region Character
and the gray values Image with the OCR classifier
OCRHandle and returns the best Num classes in
Class and the corresponding confidences (probabilities) of the
classes in Confidence. Because multiple classes may be returned by
do_ocr_single_class_cnn, Character may only contain
a single region (a single character). If multiple characters should
be classified in a single call, do_ocr_multi_class_cnn must
be used.
In most cases, do_ocr_multi_class_cnn should be preferred over
do_ocr_single_class_cnn, since do_ocr_multi_class_cnn is
typically faster than a loop with do_ocr_single_class_cnn.
Furthermore, the resulting confidences can be interpreted as
probabilities. They indicate the uncertainty of a character’s classification
result. The operator do_ocr_single_class_cnn has to be used if the
nth-best class has to be examined explicitly for n greater than one.
A string of the number
\('\backslash 032'\) (alternatively
displayed as \('\backslash 0x1A'\)) in
Class signifies that the region has been classified as rejection
class.
Execution information🔗
Execution information
-
Multithreading type: reentrant (runs in parallel with non-exclusive operators).
-
Multithreading scope: global (may be called from any thread).
-
Processed without parallelization.
Parameters🔗
Character (input_object) region → object
Character to be recognized.
Image (input_object) singlechannelimage → object (byte / uint2)
Gray values of the character.
OCRHandle (input_control) ocr_cnn → (handle)
Handle of the OCR classifier.
Num (input_control) integer-array → (integer)
Number of best classes to determine.
Default: 1
Suggested values: 1, 2, 3, 4, 5
Class (output_control) string(-array) → (string)
Result of classifying the character with the CNN.
Confidence (output_control) real(-array) → (real)
Confidence(s) of the class(es) of the character.
Example🔗
(HDevelop)
read_image(Image, 'bottle2')
OffsetRow := 100
OffsetCol := 108
read_ocr_class_cnn('Universal_0-9_NoRej.occ', OCRHandle)
* Select each digit and use do_ocr_single_class_cnn to apply OCR
gen_rectangle1(ROI_Date, OffsetRow, OffsetCol, OffsetRow, OffsetCol)
for I := 1 to 6 by 1
Offset := I % 2 * 10
smallest_rectangle1(ROI_Date, Row1, Col1, Row2, Col2)
gen_rectangle1(ROI_Date, OffsetRow, Offset + Col2, OffsetRow + 42, \
Col2 + Offset + 31)
do_ocr_single_class_cnn(ROI_Date, Image, OCRHandle, 1, Class, Confidence)
endfor
Result🔗
If the parameters are valid, the operator
do_ocr_single_class_cnn returns the value 2 (H_MSG_TRUE). If
necessary, an exception is raised.
Combinations with other operators🔗
Module🔗
OCR/OCV