Training of an OCR classifier by the input of regions.
The operator ::traind_ocr_class_box trains the classifier directly via the input of regions in an image. Any number of regions of an image can be passed. For each character (region) in Character the corresponding name (class) Class must be specified. The grayvalues are passed in Image. When the procedure has finished the parameter AvgConfidence provides information about the success of the training: It contains the average confidence of the trained characters measured by a re-classification. The confidence of mismatched characters is set to 0 (thus, the average confidence will be decreased significantly).
|
Character (input_object) |
region(-array) -> Hobject: HRegion(Array) |
| Characters to be trained. | |
|
Image (input_object) |
image -> Hobject: HImage |
| Grayvalues for the characters. | |
|
OcrHandle (input_control) |
ocr -> HTuple.long |
| ID of the desired OCR-classifier. | |
|
Class (input_control) |
string(-array) -> HTuple.char * |
| Class (name) of the characters. | |
| Default value: ''a'' | |
|
AvgConfidence (output_control) |
real -> double * |
| Average confidence during a re-classification of the trained characters. | |
char name[128];
long orc_handle;
read_ocr("testnet",&orc_handle);
read_image(&Image,"character.tiff");
bin_threshold(Image,&Dark);
connection(Dark,&Character);
count_obj(Character,&num);
open_window(0,0,-1,-1,0,"","",&WindowHandle);
set_color(WindowHandle,"red");
for (i=0; i<num; i++) {
select_obj(Character,&SingleCharacter,i);
clear_window(WindowHandle);
disp_region(SingleCharacter,WindowHandle);
printf("class of character %d ?\n",i);
scanf("%s\n",name);
traind_ocr_class_box(SingleCharacter,Image,OcrHandle,name,&AvgConfidence);
}
If the parameters are correct, the operator ::traind_ocr_class_box returns the value H_MSG_TRUE. Otherwise an exception will be raised.
::create_ocr_class_box, ::read_ocr
::traind_ocr_class_box, ::write_ocr, ::do_ocr_multi, ::do_ocr_single
Optical character recognition
