Differenze tra le versioni di "Eseguire una scansione e riconoscimento caratteri (OCR) con l'API di Sane, Tesseract e Leptonica"

Versione attuale delle 19:24, 1 gen 2024

Usando le risorse esterne del API di Sane, Tesseract e Leptonica è possibile eseguire una scansione con la Scanner e il successivo riconoscimento caratteri (OCR) di un documento testuale.

In particolare con le funzioni esterne della libreria libsane si eseguirà la scansione attraverso uno scanner; con le funzioni esterne della libreria liblept si caricherà l'immagine ottenuta dalla scansione del documento, e con le funzioni esterne della libreria libtesseract si eseguirà il riconoscimento ottico dei caratteri presenti nell'immagine caricata.

E' necessario avere installate nel sistema e richiamare in Gambas le seguenti librerie condivise:
"libsane.so.1.2.1 ", "libtesseract.so.4.0.1 " e "liblept.so.5.0.4 "

Mostriamo un esempio pratico:

Private Const BUFFER As Integer = 65536
Private Const MODUS As String = "Color"   ' Color  -  Gray  -  Lineart
Private RISOLUZIONE As Integer = 300
Private Const OCR_BOOL As Boolean = True

Library "libsane:1.2.1"

Public Struct SANE_Option_Descriptor
  name As Pointer
  title As Pointer
  desc As Pointer
  type As Integer
  unit As Integer
  size As Integer
  cap As Integer
  constraint_type As Integer
  union As Pointer
End Struct

Public Struct SANE_Parameters
  sformat As Integer
  last_frame As Integer
  bytes_per_line As Integer
  pixels_per_line As Integer
  lines As Integer
  depth As Integer
End Struct

Private Enum SANE_ACTION_GET_VALUE = 0, SANE_ACTION_SET_VALUE, SANE_ACTION_SET_AUTO

' SANE_Status sane_init (SANE_Int * version_code, SANE_Auth_Callback authorize)
' This function must be called before any other SANE function can be called.
Private Extern sane_init(version_code As Pointer, authorize As Pointer) As Integer

' SANE_Status sane_get_devices (const SANE_Device *** device_list, SANE_Bool local_only)
' To query the list of devices that are available.
Private Extern sane_get_devices(device_list As Pointer, local_only As Boolean) As Integer

' SANE_Status sane_open (SANE_String_Const devicename, SANE_Handle * handle)
' To establish a connection to a particular device.
Private Extern sane_open(devicename As String, handle As Pointer) As Integer

' SANE_Status sane_control_option (SANE_Handle handle, SANE_Int option, SANE_Action action, void *value, SANE_Int * info)
' To set or inquire the current value of option number n of the device represented by handle h.
Private Extern sane_control_option(handle As Pointer, option As Integer, action As Integer, value As Pointer, info As Pointer) As Integer

' const SANE_Option_Descriptor * sane_get_option_descriptor (SANE_Handle handle, SANE_Int option)
' To access option descriptors.
Private Extern sane_get_option_descriptor(handle As Pointer, option As Integer) As Pointer

' SANE_Status sane_get_parameters (SANE_Handle handle, SANE_Parameters * params)
' To obtain the current scan parameters.
Private Extern sane_get_parameters(handle As Pointer, params As SANE_Parameters) As Integer

' SANE_Status sane_start (SANE_Handle handle)
' Initiates aquisition of an image from the device represented by handle h.
Private Extern sane_start(handle As Pointer) As Integer

' SANE_Status sane_read (SANE_Handle handle, SANE_Byte * data, SANE_Int max_length, SANE_Int * length)
' Reads image data from the device represented by handle h.
Private Extern sane_read(handle As Pointer, data As Pointer, max_length As Integer, length As Pointer) As Integer

' void sane_cancel (SANE_Handle handle)
' Cancels the currently pending operation of the device.
Private Extern sane_cancel(handle As Pointer)

' void sane_close (SANE_Handle h)
' Terminates the association between the device handle passed in argument h and the device it represents.
Private Extern sane_close(h As Pointer)

' void sane_exit (void)
' To terminate use of a backend.
Private Extern sane_exit()


Public Sub Main()
 
 Dim i, n, maxlun, lun As Integer
 Dim dev, hnd, md, buf, plun As Pointer
 Dim nom As String
 Dim sod As SANE_Option_Descriptor
 Dim par As New SANE_Parameters
 Dim fl As File
 
 Write "\e[5mRestare in attesa...\e[0m"
 Flush
 
 i = sane_init(0, 0)
 If i > 0 Then GestErrore("sane_init()")

 i = sane_get_devices(VarPtr(dev), False)
 If i > 0 Then GestErrore("sane_get_devices()")
 
 While Pointer@(dev + i) > 0
   nom = String@(Pointer@(Pointer@(dev + i)))
   i += 8
 Wend
 Write "\rDispositivo:     \e[31m" & nom

 i = sane_open(nom, VarPtr(hnd))
 If i > 0 Then
   GestErrore("sane_open()")
   saneUscita(hnd)
 Endif

 i = sane_control_option(hnd, 0, SANE_ACTION_GET_VALUE, VarPtr(n), 0)
 If i > 0 Then
   GestErrore("sane_control_option()")
   saneUscita(hnd)
 Endif
 
 Print "\n\n\e[0mPossibili opzioni:\e[32m"
 For i = 0 To n - 1
   sod = sane_get_option_descriptor(hnd, i)
   If IsNull(String@(sod.name)) Then Continue
   Print "   " & String@(sod.name)
   Select Case String@(sod.name)
     Case "mode"                    ' Color  -  Gray  -  Lineart
       md = Alloc(MODUS)
       sane_control_option(hnd, i, SANE_ACTION_SET_VALUE, md, 0)
     Case "resolution"
       sane_control_option(hnd, i, SANE_ACTION_SET_VALUE, VarPtr(RISOLUZIONE), 0)
   End Select
 Next

 i = sane_get_parameters(hnd, par)
 If i > 0 Then
   GestErrore("sane_get_parameters()")
   saneUscita(hnd)
 Endif

 With par
   Dim px As Byte = .bytes_per_line / .pixels_per_line
   Dim rgb As String = " byte)"
   If px == 3 Then rgb = " byte = RGB)"
   Print "\n\e[0mColonne:          "; .pixels_per_line; " (1 colonna = 1 pixel = "; px; RGB
   Print "Righe:            "; .lines
   Print "Byte per riga:    "; .bytes_per_line; " (colonne x "; px; Left(rgb, 5); ")"
   Print "Bit per campione: "; .depth
   Print "Modalità:        "; MODUS
   Print "Risoluzione:     "; RISOLUZIONE; " dpi\n"
 End With
 
 i = sane_start(hnd)
 If i > 0 Then
   GestErrore(hnd, "sane_start()")
   saneUscita(hnd)
 Endif

 buf = Alloc(SizeOf(gb.Byte), BUFFER)
 plun = Alloc(SizeOf(gb.Integer), 1)
 maxlun = BUFFER

 fl = Open "/tmp/imm" For Create

 Repeat
   i = sane_read(hnd, buf, maxlun, plun)
   Select Case i
     Case 1
       GestErrore("sane_read()")
       saneUscita(hnd)
     Case 3 To 4
       GestErrore("sane_read()")
       saneUscita(hnd)
     Case 6 To 11
       GestErrore("sane_read()")
       saneUscita(hnd)
   End Select
   lun = Int@(plun)
   Write #fl, buf, lun
 Until lun = 0

 CreaFileImmagine(par)
 
' Libera la memoria precedentemente allocata:
 fl.Close
 Free(md)
 Free(plun)
 Free(buf)
 saneUscita(hnd)

' Chiama la funzione per il riconoscimento OCR:
 If OCR_BOOL = True Then Ocr()
 
End

 
Private Procedure GestErrore(fnz As String)

 Error.Raise("Errore alla funzione: " & fnz)

End


Private Procedure saneUscita(handle As Pointer)

 sane_cancel(handle)
 sane_close(handle)
 sane_exit()

End

Private Procedure CreaFileImmagine(prmt As SANE_Parameters)

 Dim s, finale, mg As String
 Dim pro, p4, colonne, righe As Integer
 Dim bBN As Byte

 pro = prmt.lines * prmt.bytes_per_line
 s = File.Load("/tmp/imm")

 Print "Dati grezzi: "; Len(s); " byte (Colonne x Righe x 3 byte_RGB)"

 colonne = prmt.pixels_per_line
 righe = prmt.lines

 Select Case MODUS
   Case "Color"
     mg = "P6 "
' Se il numero dei dati grezzi è inferiore a (Colonne x Righe) x 3, allora colma la differenza:
     If Len(s) < ((pro / 8) * 3) Then s = s & String(((pro) * 3) - Len(s), Chr(255))
' Se invece il numero dei dati grezzi è maggiore di (Colonne x Righe) x 3, allora colma la differenza, ma tagliando i dati grezzi dalla fine:
     If Len(s) > ((pro) * 3) Then s = Left(s, pro * 3)
   Case "Gray"
     mg = "P5 "
' Se il numero dei dati grezzi è inferiore a Colonne x Righe, allora colma la differenza:
     If Len(s) < (pro) Then s = s & String((pro) - Len(s), Chr(255))
' Se invece il numero dei dati grezzi è maggiore di Colonne x Righe, allora colma la differenza, ma tagliando i dati grezzi dalla fine:
     If Len(s) > (pro) Then s = Left(s, pro)
   Case "Lineart"
     mg = "P4 "
' Se il numero dei dati grezzi è inferiore a (Colonne x Righe) / 8, allora colma la differenza:
     If Len(s) < (pro / 8) Then
       p4 = pro \ 8
       s = s & String$(p4 - Len(s), Chr(255))
     Endif
' Se il numero dei dati grezzi è maggiore di (Colonne x Righe) / 8, allora colma la differenza:
     If Len(s) > (pro / 8) Then
       For bBN = 1 To 8
         If (Len(s) * 8) Mod (prmt.lines + bBN) == 0
           colonne = prmt.lines + bBN
           righe = Len(s) * 8 / prmt.lines
         Endif
       Next
     Endif
 End Select
' Quindi, crea l'intero file .PNM finale, costituito da:
' l'header:
'   - numero Magico + spazio;
'   - numero colonne;
'   - spazio;
'   - numero righe;
'   - spazio;
'   - profondità colore;
'   - chiusura con "salto di riga";
' i dati grezzi dell'immagine scansita;
' le informazioni e commento finale:
 finale = mg & CStr(colonne) & Chr(&20) & CStr(righe) & " 255" & Chr(10) & s & Chr(10) &
          "# file creato da " & Application.Name & " - " & CStr(Now)
' ...e lo salva:
 File.Save("/tmp/scansione.pnm", finale)
 
End


Public Struct Pix
  w As Integer
  h As Integer
  d As Integer
  wpl As Integer
  refcount As Integer
  xres As Integer
  yres As Integer
  informat As Integer
  text As Pointer
  colormap As Pointer
  data As Pointer
End Struct

Library "libtesseract:4.0.1"

' TessBaseAPI* TessBaseAPICreate()
Private Extern TessBaseAPICreate() As Pointer

' int TessBaseAPIInit3(TessBaseAPI* handle, const char* datapath, const char* language)
Private Extern TessBaseAPIInit3(handle As Pointer, datapath As String, language As String) As Integer

' void TessBaseAPISetImage2(TessBaseAPI* handle, struct Pix* pix)
Private Extern TessBaseAPISetImage2(handle As Pointer, pix As Pix)

' void TessBaseAPISetSourceResolution(TessBaseAPI* handle, int ppi)
Private Extern TessBaseAPISetSourceResolution(handle As Pointer, ppi As Integer)

' int TessBaseAPIRecognize(TessBaseAPI* handle, ETEXT_DESC* monitor)
Private Extern TessBaseAPIRecognize(handle As Pointer, monitor As Pointer) As Integer

' char* TessBaseAPIGetUTF8Text(TessBaseAPI* handle)
Private Extern TessBaseAPIGetUTF8Text(handle As Pointer) As String

' void TessBaseAPIEnd(TessBaseAPI* handle)
Private Extern TessBaseAPIEnd(handle As Pointer)

' void TessBaseAPIDelete(TessBaseAPI* handle)
Private Extern TessBaseAPIDelete(handle As Pointer)


Library "liblept:5.0.4"

' PIX * pixRead (const char *filename)
Private Extern pixRead(filename As String) As Pix

' void pixDestroy (PIX **ppix)
Private Extern pixDestroy(ppix As Pointer)
 
 
Private Procedure Ocr()
 
 Dim err As Integer
 Dim tes, p As Pointer 
 Dim pis As Pix
 Dim s As String
 
 pis = pixRead("/tmp/scansione.pnm")
 
 tes = TessBaseAPICreate()
 If tes == 0 Then Error.Raise("Errore !")
 
 err = TessBaseAPIInit3(tes, Null, "eng")
 If err <> 0 Then Error.Raise("Errore !")
 
 TessBaseAPISetImage2(tes, pis)
 Write "Riconoscimento caratteri.\n\e[5m  Restare in attesa...\e[0m\r"
 Flush
 
 TessBaseAPISetSourceResolution(tes, RISOLUZIONE)
 
 err = TessBaseAPIRecognize(tes, 0)
 If err <> 0 Then Error.Raise("Errore !")
 
' Scansione OCR:
 s = TessBaseAPIGetUTF8Text(tes)
 Print Space(Len("  Restare in attesa..."))
 Print
 Print s
 
 TessBaseAPIEnd(tes)
 TessBaseAPIDelete(tes)
 pixDestroy(Object.Address(pis) + (SizeOf(gb.Pointer) * 3))
 
End