Fixed typos and improved comments

git-svn-id: https://tesseract-ocr.googlecode.com/svn/trunk@753 d0cd1f9f-072b-0410-8dd7-cf729c803f20

README

@@ -26,6 +26,7 @@ Dependencies and Licenses
 
 Leptonica is required. (www.leptonica.com). Tesseract no longer compiles
 without Leptonica.
+Libtiff is no longer required as a direct dependency.
 
 
 Installing and Running Tesseract

ccmain/paragraphs_internal.h

@@ -277,7 +277,7 @@ void RecomputeMarginsAndClearHypotheses(
     GenericVector<RowScratchRegisters> *rows, int start, int end,
     int percentile);
 
-// Return the minimum inter-word space in rows[row_start, row_end).
+// Return the median inter-word space in rows[row_start, row_end).
 int InterwordSpace(const GenericVector<RowScratchRegisters> &rows,
                    int row_start, int row_end);
 

ccstruct/ratngs.h

@@ -138,7 +138,16 @@ class BLOB_CHOICE: public ELIST_LINK
   UNICHAR_ID unichar_id_;          // unichar id
   inT16 fontinfo_id_;              // char font information
   inT16 fontinfo_id2_;             // 2nd choice font information
+  // Rating is the classifier distance weighted by the length of the outline
+  // in the blob. In terms of probability, classifier distance is -klog p such
+  // that the resulting distance is in the range [0, 1] and then
+  // rating = w (-k log p) where w is the weight for the length of the outline.
+  // Sums of ratings may be compared meaningfully for words of different
+  // segmentation.
   float rating_;                  // size related
+  // Certainty is a number in [-20, 0] indicating the classifier certainty
+  // of the choice. In terms of probability, certainty = 20 (k log p) where
+  // k is defined as above to normalize -klog p to the range [0, 1].
   float certainty_;               // absolute
   int script_id_;
   // Stores language model information about this BLOB_CHOICE. Used during
@@ -408,7 +417,9 @@ class WERD_CHOICE {
   char *fragment_lengths_;   // number of fragments in each unichar
   int reserved_;             // size of the above arrays
   int length_;               // word length
+  // Rating is the sum of the ratings of the individual blobs in the word.
   float rating_;             // size related
+  // certainty is the min (worst) certainty of the individual blobs in the word.
   float certainty_;          // absolute
   uinT8 permuter_;           // permuter code
   bool fragment_mark_;       // if true, indicates that this choice

classify/adaptmatch.cpp

@@ -181,12 +181,10 @@ void Classify::AdaptiveClassifier(TBLOB *Blob,
                                   CLASS_PRUNER_RESULTS CPResults) {
   assert(Choices != NULL);
   ADAPT_RESULTS *Results = new ADAPT_RESULTS();
+  Results->Initialize();
 
   if (AdaptedTemplates == NULL)
     AdaptedTemplates = NewAdaptedTemplates (true);
-
-  Results->Initialize();
-
   DoAdaptiveMatch(Blob, denorm, Results);
   if (CPResults != NULL)
     memcpy(CPResults, Results->CPResults,
@@ -903,7 +901,14 @@ int Classify::AdaptableWord(TWERD *Word,
       BestChoiceLength > 0 &&
       BestChoiceLength == Word->NumBlobs() &&
       BestChoiceLength <= MAX_ADAPTABLE_WERD_SIZE &&
+      // This basically ensures that the word is at least a dictionary match
+      // (freq word, user word, system dawg word, etc).
+      // Since all the other adjustments will make adjust factor higher
+      // than higher than adaptable_score=1.1+0.05=1.15
+      // Since these are other flags that ensure that the word is dict word,
+      // this check could be at times redundant.
       getDict().CurrentBestChoiceAdjustFactor() <= adaptable_score &&
+      // Make sure that alternative choices are not dictionary words.
       getDict().AlternativeChoicesWorseThan(adaptable_score) &&
       getDict().CurrentBestChoiceIs(BestChoiceWord);
 }
@@ -2487,18 +2492,12 @@ void Classify::RemoveBadMatches(ADAPT_RESULTS *Results) {
 
 /*----------------------------------------------------------------------------*/
 /**
- * This routine steps thru each matching class in Results
+ * This routine discards extra digits or punctuation from the results.
- * and removes it from the match list if its rating
+ * We keep only the top 2 punctuation answers and the top 1 digit answer if
- * is worse than the BestRating plus a pad.  In other words,
+ * present.
- * all good matches get moved to the front of the classes
- * array.
  *
  * @param Results contains matches to be filtered
  *
- * Globals:
- * - matcher_bad_match_pad defines a "bad match"
- *
- * @note Exceptions: none
  * @note History: Tue Mar 12 13:51:03 1991, DSJ, Created.
  */
 void Classify::RemoveExtraPuncs(ADAPT_RESULTS *Results) {

classify/intproto.cpp

@@ -1846,7 +1846,6 @@ void RenderIntFeature(ScrollView *window, const INT_FEATURE_STRUCT* Feature,
  * proto from the class description and adds a rendering of
  * the proto onto the ShapeList.
  *
- * @param ShapeList   shape list to append proto rendering onto
  * @param Class   class that proto is contained in
  * @param ProtoId   id of proto to be rendered
  * @param color   color to render proto in