enable to select pdf compression type and jpeg quality (fix issue 1263)

git-svn-id: https://tesseract-ocr.googlecode.com/svn/trunk@1134 d0cd1f9f-072b-0410-8dd7-cf729c803f20

Author: zdenop

api/pdfrenderer.cpp

@@ -519,7 +519,7 @@ bool TessPDFRenderer::fileToPDFObj(char *filename, long int objnum,
   return true;
 }
 
-bool TessPDFRenderer::pixToPDFObj(Pix *pix, long int objnum,
+bool TessPDFRenderer::pixToPDFObj(TessBaseAPI* api, Pix *pix, long int objnum,
                                   char **pdf_object,
                                   long int *pdf_object_size) {
   if (!pdf_object_size || !pdf_object)
@@ -531,11 +531,17 @@ bool TessPDFRenderer::pixToPDFObj(Pix *pix, long int objnum,
   char b2[kBasicBufSize];
   L_COMP_DATA *cid;
   int encoding_type;
-  const int kJpegQuality = 85;
-  if (selectDefaultPdfEncoding(pix, &encoding_type) != 0)
-    return false;
+  int kJpegQuality;
+  api->GetIntVariable("tessedit_pdf_jpg_quality", &kJpegQuality);
+  api->GetIntVariable("tessedit_pdf_compression", &encoding_type);
+  if (encoding_type == 0 || encoding_type > 3) {
+    if (selectDefaultPdfEncoding(pix, &encoding_type) != 0)
+      return false;
+  }
+
   if (pixGenerateCIData(pix, encoding_type, kJpegQuality, 0, &cid) != 0)
     return false;
+  encoding_type = cid->type;
 
   const char *filter;
   switch(encoding_type) {
@@ -677,7 +683,7 @@ bool TessPDFRenderer::AddImageHandler(TessBaseAPI* api) {
 
   char *pdf_object;
   if (!fileToPDFObj(filename, obj_, &pdf_object, &objsize)) {
-    if (!pixToPDFObj(pix, obj_, &pdf_object, &objsize)) {
+    if (!pixToPDFObj(api, pix, obj_, &pdf_object, &objsize)) {
       return false;
     }
   }

api/renderer.h

@@ -198,7 +198,8 @@ class TESS_API TessPDFRenderer : public TessResultRenderer {
   static bool fileToPDFObj(char *filename, long int objnum,
                            char **pdf_object, long int *pdf_object_size);
   // Turn a Pix into a the very best PDF object that we can.
-  static bool pixToPDFObj(Pix *pix, long int objnum,
+  static bool pixToPDFObj(tesseract::TessBaseAPI *api, Pix *pix,
+                          long int objnum,
                           char **pdf_object, long int *pdf_object_size);
 };
 

ccmain/tesseractclass.cpp

@@ -315,6 +315,16 @@ Tesseract::Tesseract()
                 "Write .html hOCR output file", this->params()),
     BOOL_MEMBER(tessedit_create_pdf, false,
                 "Write .pdf output file", this->params()),
+    INT_MEMBER(tessedit_pdf_compression, 0,
+               "Type of image compression in pdf output: "
+               "0 - autoselection (default); "
+               "1 - jpeg; "
+               "2 - G4; "
+               "3 - flat",
+               this->params()),
+    INT_MEMBER(tessedit_pdf_jpg_quality, 85,
+               "Quality level of jpeg image compression in pdf output",
+               this->params()),
     STRING_MEMBER(unrecognised_char, "|",
                   "Output char for unidentified blobs", this->params()),
     INT_MEMBER(suspect_level, 99, "Suspect marker level", this->params()),

ccmain/tesseractclass.h

@@ -908,6 +908,13 @@ class Tesseract : public Wordrec {
   BOOL_VAR_H(tessedit_write_unlv, false, "Write .unlv output file");
   BOOL_VAR_H(tessedit_create_hocr, false, "Write .html hOCR output file");
   BOOL_VAR_H(tessedit_create_pdf, false, "Write .pdf output file");
+  INT_VAR_H(tessedit_pdf_compression, 0, "Type of image encoding in pdf output:"
+            "0 - autoselection (default); "
+            "1 - jpeg; "
+            "2 - G4; "
+            "3 - flat");
+  INT_VAR_H(tessedit_pdf_jpg_quality, 85, "Quality level of jpeg image "
+            "compression in pdf output");
   STRING_VAR_H(unrecognised_char, "|",
                "Output char for unidentified blobs");
   INT_VAR_H(suspect_level, 99, "Suspect marker level");