tesseract  4.0.0-beta.1-59-g2cc4
tesseract::BaselineDetect Class Reference

#include <baselinedetect.h>

Public Member Functions

 BaselineDetect (int debug_level, const FCOORD &page_skew, TO_BLOCK_LIST *blocks)
 
 ~BaselineDetect ()
 
void ComputeStraightBaselines (bool use_box_bottoms)
 
void ComputeBaselineSplinesAndXheights (const ICOORD &page_tr, bool enable_splines, bool remove_noise, bool show_final_rows, Textord *textord)
 

Detailed Description

Definition at line 242 of file baselinedetect.h.

Constructor & Destructor Documentation

◆ BaselineDetect()

tesseract::BaselineDetect::BaselineDetect ( int  debug_level,
const FCOORD page_skew,
TO_BLOCK_LIST *  blocks 
)

Definition at line 791 of file baselinedetect.cpp.

793  : page_skew_(page_skew), debug_level_(debug_level) {
794  TO_BLOCK_IT it(blocks);
795  for (it.mark_cycle_pt(); !it.cycled_list(); it.forward()) {
796  TO_BLOCK* to_block = it.data();
797  BLOCK* block = to_block->block;
798  POLY_BLOCK* pb = block->poly_block();
799  // A note about non-text blocks.
800  // On output, non-text blocks are supposed to contain a single empty word
801  // in each incoming text line. These mark out the polygonal bounds of the
802  // block. Ideally no baselines should be required, but currently
803  // make_words crashes if a baseline and xheight are not provided, so we
804  // include non-text blocks here, but flag them for special treatment.
805  bool non_text = pb != NULL && !pb->IsText();
806  blocks_.push_back(new BaselineBlock(debug_level_, non_text, to_block));
807  }
808 }
POLY_BLOCK * poly_block() const
Definition: pdblock.h:55
bool IsText() const
Definition: polyblk.h:52
BLOCK * block
Definition: blobbox.h:773
Definition: ocrblock.h:30

◆ ~BaselineDetect()

tesseract::BaselineDetect::~BaselineDetect ( )

Definition at line 810 of file baselinedetect.cpp.

810  {
811 }

Member Function Documentation

◆ ComputeBaselineSplinesAndXheights()

void tesseract::BaselineDetect::ComputeBaselineSplinesAndXheights ( const ICOORD page_tr,
bool  enable_splines,
bool  remove_noise,
bool  show_final_rows,
Textord textord 
)

Definition at line 848 of file baselinedetect.cpp.

852  {
853  for (int i = 0; i < blocks_.size(); ++i) {
854  BaselineBlock* bl_block = blocks_[i];
855  if (enable_splines)
856  bl_block->PrepareForSplineFitting(page_tr, remove_noise);
857  bl_block->FitBaselineSplines(enable_splines, show_final_rows, textord);
858  if (show_final_rows) {
859  bl_block->DrawFinalRows(page_tr);
860  }
861  }
862 }

◆ ComputeStraightBaselines()

void tesseract::BaselineDetect::ComputeStraightBaselines ( bool  use_box_bottoms)

Definition at line 816 of file baselinedetect.cpp.

816  {
817  GenericVector<double> block_skew_angles;
818  for (int i = 0; i < blocks_.size(); ++i) {
819  BaselineBlock* bl_block = blocks_[i];
820  if (debug_level_ > 0)
821  tprintf("Fitting initial baselines...\n");
822  if (bl_block->FitBaselinesAndFindSkew(use_box_bottoms)) {
823  block_skew_angles.push_back(bl_block->skew_angle());
824  }
825  }
826  // Compute a page-wide default skew for blocks with too little information.
827  double default_block_skew = page_skew_.angle();
828  if (!block_skew_angles.empty()) {
829  default_block_skew = MedianOfCircularValues(M_PI, &block_skew_angles);
830  }
831  if (debug_level_ > 0) {
832  tprintf("Page skew angle = %g\n", default_block_skew);
833  }
834  // Set bad lines in each block to the default block skew and then force fit
835  // a linespacing model where it makes sense to do so.
836  for (int i = 0; i < blocks_.size(); ++i) {
837  BaselineBlock* bl_block = blocks_[i];
838  bl_block->ParallelizeBaselines(default_block_skew);
839  bl_block->SetupBlockParameters(); // This replaced compute_row_stats.
840  }
841 }
float angle() const
find angle
Definition: points.h:249
bool empty() const
Definition: genericvector.h:91
int push_back(T object)
#define tprintf(...)
Definition: tprintf.h:31
T MedianOfCircularValues(T modulus, GenericVector< T > *v)
Definition: linlsq.h:111

The documentation for this class was generated from the following files: