Halcon Digital Recognition instance annotation on bottle

* Key mastery of digital segmentation technology
* Bottle2.hdev:Segmenting and reading of numbers on a beer bottle
* Program without visualization: "Bottle.hdev"
*
* Step 0:preparations
* Specify the name of the the font to the reading the date on the bottle.
* It's easiest to use the pre-trained font industrial_0-9. If you have run the
* Program Bottlet.hdev in this directory, you can activate the second line
* To use the font trained.
FontName: = ' industrial_0-9 '
* FontName: = ' bottle '
*
* Step 1:segmentation
Dev_update_window (' off ')
Read_image (Bottle, ' bottle2 ')
Get_image_size (Bottle, Width, Height)
Dev_close_window ()
Dev_open_window (0, 0, 2*width, 2*height, ' black ', windowid)
Set_display_font (Windowid, ' mono ', ' true ', ' false ')
Dev_display (Bottle)
Disp_continue_message (Windowid, ' black ', ' true ')
Stop ()
* Binary, select the darker part
Threshold (Bottle, rawsegmentation, 0, 95)
* Fill up some small holes in the middle
Fill_up_shape (rawsegmentation, removednoise, ' area ', 1, 5)
* Open operation, remove the small impurity image

Opening_circle (Removednoise, thickstructures, 2.5)

Dev_display (Bottle)
Dev_set_color (' green ')
Dev_display (Thickstructures)
Disp_continue_message (Windowid, ' black ', ' true ')
Stop ()
Dev_set_draw (' margin ')
Dev_set_line_width (3)
Dev_display (Bottle)
Dev_display (Thickstructures)
Disp_continue_message (Windowid, ' black ', ' true ')
Stop ()
* Fill area, inside hole sealed up, handle edge
Fill_up (Thickstructures, Solid)
Dev_display (Bottle)
Dev_clear_window ()
Dev_display (Solid)
Disp_continue_message (Windowid, ' black ', ' true ')

Stop ()

* Rectangular open operation, separate the connected small area
Opening_rectangle1 (Solid, Cut, 1, 7)
Dev_display (Bottle)
Dev_display (Cut)
Disp_continue_message (Windowid, ' black ', ' true ')

Stop ()

* Calculate connected areas
Connection (Cut, Connectedpatterns)
* Intersection
Dev_clear_window ()
Dev_display (Connectedpatterns)
Dev_clear_window ()
Dev_display (Thickstructures)
Intersection (Connectedpatterns, Thickstructures, Numbercandidates)
Dev_clear_window ()
Dev_display (Numbercandidates)


Dev_set_colored (12)
Dev_display (Bottle)
Dev_display (Numbercandidates)
Disp_continue_message (Windowid, ' black ', ' true ')

Stop ()

* Select Area

Select_shape (Numbercandidates, Numbers, ' area ', ' and ', 300, 9999)

* Queued Sort_region (Numbers, finalnumbers, ' First_point ', ' true ', ' column ')
Dev_set_color (' Red ')
Dev_display (Bottle)
Dev_display (Numbers)
Dev_set_shape (' Rectangle1 ')
Dev_set_color (' green ')
Dev_display (Finalnumbers)
Dev_set_shape (' original ')
Dev_set_line_width (1)
Dev_set_draw (' Fill ')
*
* step2:reading
* Read the well-trained classifier. This is supposed to be a neural network and stuff like that.
READ_OCR_CLASS_MLP (FontName, Ocrhandle)
* Identification
DO_OCR_MULTI_CLASS_MLP (Finalnumbers, Bottle, Ocrhandle, Recnum, Confidence)
* Computing Center Point
Area_center (finalnumbers, area, Row, Colnum)
Set_display_font (Windowid, +, ' mono ', ' true ', ' false ')
For I: = 0 to | Recnum|-1 by 1
Disp_message (Windowid, recnum[i], ' image ', A, colnum[i]-3, ' green ', ' false ')
ENDfor

* Display effect

CLEAR_OCR_CLASS_MLP (Ocrhandle)
Dev_update_window (' on ')