/ocr/ocrservice/jni/hydrogen/src/hydrogentextdetector.h

  1/*
  2 * Copyright 2010, Google Inc.
  3 *
  4 * Licensed under the Apache License, Version 2.0 (the "License");
  5 * you may not use this file except in compliance with the License.
  6 * You may obtain a copy of the License at
  7 *
  8 *     http://www.apache.org/licenses/LICENSE-2.0
  9 *
 10 * Unless required by applicable law or agreed to in writing, software
 11 * distributed under the License is distributed on an "AS IS" BASIS,
 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 13 * See the License for the specific language governing permissions and
 14 * limitations under the License.
 15 */
 16
 17#ifndef HYDROGEN_HYDROGENTEXTDETECTOR_H_
 18#define HYDROGEN_HYDROGENTEXTDETECTOR_H_
 19
 20#include "leptonica.h"
 21
 22class HydrogenTextDetector {
 23public:
 24  HydrogenTextDetector();
 25
 26  ~HydrogenTextDetector();
 27
 28  struct TextDetectorParameters {
 29    bool debug;
 30    char out_dir[255];
 31
 32    // Edge-based thresholding
 33    l_int32 edge_tile_x;
 34    l_int32 edge_tile_y;
 35    l_int32 edge_thresh;
 36    l_int32 edge_avg_thresh;
 37
 38    // Skew angle correction
 39    bool skew_enabled;
 40    l_float32 skew_min_angle;
 41    l_float32 skew_sweep_range;
 42    l_float32 skew_sweep_delta;
 43    l_int32 skew_sweep_reduction;
 44    l_int32 skew_search_reduction;
 45    l_float32 skew_search_min_delta;
 46
 47    // Singleton filter
 48    l_float32 single_min_aspect;
 49    l_float32 single_max_aspect;
 50    l_int32 single_min_area;
 51    l_float32 single_min_density;
 52
 53    // Quick pair filter
 54    l_float32 pair_h_ratio;
 55    l_float32 pair_d_ratio;
 56    l_float32 pair_h_dist_ratio;
 57    l_float32 pair_v_dist_ratio;
 58    l_float32 pair_h_shared;
 59
 60    // Cluster pair filter
 61    l_int32 cluster_width_spacing;
 62    l_float32 cluster_shared_edge;
 63    l_float32 cluster_h_ratio;
 64
 65    // Finalized cluster filter
 66    l_int32 cluster_min_blobs;
 67    l_float32 cluster_min_aspect;
 68    l_float32 cluster_min_fdr;
 69    l_int32 cluster_min_edge;
 70    l_int32 cluster_min_edge_avg;
 71
 72    TextDetectorParameters()
 73        : debug(false),
 74          edge_tile_x(32),
 75          edge_tile_y(64),
 76          edge_thresh(64),
 77          edge_avg_thresh(4),
 78          skew_enabled(true),
 79          skew_min_angle(1.0),
 80          skew_sweep_range(30.0),
 81          skew_sweep_delta(5.0),
 82          skew_sweep_reduction(8),
 83          skew_search_reduction(4),
 84          skew_search_min_delta(0.01),
 85          single_min_aspect(0.1),
 86          single_max_aspect(4.0),
 87          single_min_area(4),
 88          single_min_density(0.2),
 89          pair_h_ratio(1.0),
 90          pair_d_ratio(1.5),
 91          pair_h_dist_ratio(2.0),
 92          pair_v_dist_ratio(0.25),
 93          pair_h_shared(0.25),
 94          cluster_width_spacing(2),
 95          cluster_shared_edge(0.5),
 96          cluster_h_ratio(1.0),
 97          cluster_min_blobs(5),
 98          cluster_min_aspect(2),
 99          cluster_min_fdr(2.5),
100          cluster_min_edge(32),
101          cluster_min_edge_avg(1)
102          {
103    }
104  };
105
106  // Function to set the original source image
107  void SetSourceImage(PIX *);
108
109  // Main text detection function
110  void DetectText();
111
112  // Clear recognition results between calls
113  void Clear();
114
115  // Function to return text area clippings of the original image
116  PIXA *GetImageAreas();
117
118  // Function to return binarized text areas
119  PIXA *GetTextAreas();
120
121  // Function to return text area confidences
122  NUMA *GetTextConfs();
123
124  // Function to return detected skew angle
125  l_float32 GetSkewAngle();
126
127  // Function to return the original source image
128  PIX *GetSourceImage();
129
130  TextDetectorParameters *GetMutableParameters();
131
132private:
133  TextDetectorParameters parameters_;
134
135  // Source image
136  PIX *pixs_;
137  // Detected text areas
138  PIXA *text_areas_;
139  // Confidences of detected text areas
140  NUMA *text_confs_;
141  // Detected skew angle
142  l_float32 skew_angle_;
143
144  // Function to extract text areas from a PIX
145  PIXA *ExtractTextRegions(PIX *pix8, PIX *edges, NUMA **pconfs);
146
147  // Function to detect and fix text skew
148  PIX *DetectAndFixSkew(PIX *pixs);
149};
150
151#endif /* HYDROGEN_HYDROGENTEXTDETECTOR_H_ */