Merge pull request #124 from ws233/master

Renamed `confidencesByIteratorLevel` to `recognizedBlocksByIteratorLevel` and removed `characterBoxes`. Fixes #109.

Author: kevincon

TesseractOCR/G8Tesseract.h

@@ -155,14 +155,6 @@
  */
 @property (nonatomic, readonly) CGFloat deskewAngle;
 
-/**
- *  An array of `G8RecognizedBlock`'s representing the characters recognized
- *  in the target image, including the bounding boxes for each character. See
- *  G8RecognizedBlock.h for more information about the available fields for 
- *  this data structure.
- */
-@property (nonatomic, readonly) NSArray *characterBoxes;
-
 /**
  *  An array of arrays, where each subarray contains `G8RecognizedBlock`'s
  *  representing the choices Tesseract considered for each symbol in the target
@@ -175,18 +167,20 @@
 @property (nonatomic, readonly) NSArray *characterChoices;
 
 /**
- *  Retrieve Tesseract's confidences for its recognition result based on a
- *  provided resolution.
+ *  Retrieve Tesseract's recognition result based on a provided resolution.
+ *  For, example for the pageIteratorLevel == G8PageIteratorLevelSymbol it returns 
+ *  an array of `G8RecognizedBlock`'s representing the characters recognized
+ *  in the target image, including the bounding boxes for each character.
  *
  *  @param pageIteratorLevel A `G8PageIteratorLevel` representing the resolution
  *                           of interest. See G8Constants.h for the available 
  *                           resolution options.
  *
- *  @return An array of `G8RecognizedBlock`'s, each containing a confidence 
- *          value for the text it represents. See G8RecognizedBlock.h for more
+ *  @return An array of `G8RecognizedBlock`'s, each containing a confidence
+ *          value and a bounding box for the text it represents. See G8RecognizedBlock.h for more
  *          information about the available fields for this data structure.
  */
-- (NSArray *)confidencesByIteratorLevel:(G8PageIteratorLevel)pageIteratorLevel;
+- (NSArray *)recognizedBlocksByIteratorLevel:(G8PageIteratorLevel)pageIteratorLevel;
 
 #pragma mark - Debug methods
 

TesseractOCR/G8Tesseract.mm

@@ -500,44 +500,6 @@ - (void)analyzeLayout
     self.layoutAnalysed = YES;
 }
 
-- (NSArray *)characterBoxes
-{
-    NSMutableArray *recognizedTextBoxes = [[NSMutableArray alloc] init];
-
-    //  Get box info
-    char *boxText = _tesseract->GetBoxText(0);
-    if (boxText == NULL) {
-        NSLog(@"No boxes recognized. Check that -[Tesseract setImage:] is passed an image bigger than 0x0.");
-        return nil;
-    }
-
-    NSString *stringBoxes = [NSString stringWithUTF8String:boxText];
-    delete[] boxText;
-
-    NSArray *arrayOfStringBoxes = [stringBoxes componentsSeparatedByString:@"\n"];
-    for (NSString *stringBox in arrayOfStringBoxes) {
-        //  A stringBox is of the format "c L B R T p"
-        //  (L, T) is the top left corner of the box, and (R, B) is the bottom right corner
-        //  Tesseract has (0, 0) in the bottom left corner and UIKit has (0, 0) in the top left corner
-        //  Need to flip to work with UIKit
-        //  c is the recognized character and p is the page it is recognized on
-        NSArray *boxComponents = [stringBox componentsSeparatedByString:@" "];
-        if (boxComponents.count >= 6) {
-            CGFloat x = [boxComponents[1] floatValue];
-            CGFloat y = self.imageSize.height - [boxComponents[4] floatValue];
-            CGFloat width = [boxComponents[3] floatValue] - [boxComponents[1] floatValue];
-            CGFloat height = [boxComponents[4] floatValue] - [boxComponents[2] floatValue];
-            CGRect box = [self normalizedRectForX:x y:y width:width height:height];
-
-            G8RecognizedBlock *block = [[G8RecognizedBlock alloc] initWithText:boxComponents[0]
-                                                                   boundingBox:box
-                                                                    confidence:0.0f
-                                                                         level:G8PageIteratorLevelBlock];
-            [recognizedTextBoxes addObject:block];
-        }
-    }
-    return [recognizedTextBoxes copy];
-}
 
 - (CGRect)normalizedRectForX:(CGFloat)x y:(CGFloat)y width:(CGFloat)width height:(CGFloat)height
 {
@@ -557,8 +519,9 @@ - (G8RecognizedBlock *)blockFromIterator:(tesseract::ResultIterator *)iterator
     const char *word = iterator->GetUTF8Text(level);
     if (word != NULL) {
         // BoundingBox parameters are (Left Top Right Bottom).
-        // See comment in characterBoxes() for information on the coordinate
-        // system, and changes being made.
+        //  (L, T) is the top left corner of the box, and (R, B) is the bottom right corner
+        //  Tesseract has (0, 0) in the bottom left corner and UIKit has (0, 0) in the top left corner
+        //  Need to flip to work with UIKit
         int x1, y1, x2, y2;
         iterator->BoundingBox(level, &x1, &y1, &x2, &y2);
 
@@ -614,7 +577,7 @@ - (NSArray *)characterChoices
     return [array copy];
 }
 
-- (NSArray *)confidencesByIteratorLevel:(G8PageIteratorLevel)pageIteratorLevel
+- (NSArray *)recognizedBlocksByIteratorLevel:(G8PageIteratorLevel)pageIteratorLevel
 {
     tesseract::PageIteratorLevel level = (tesseract::PageIteratorLevel)pageIteratorLevel;
 

TestsProject/TestsProjectTests/RecognitionTests.m

@@ -111,7 +111,7 @@
     it(@"Should provide confidences", ^{
         [helper recognizeImage];
 
-        NSArray *confidences = [helper.tesseract confidencesByIteratorLevel:G8PageIteratorLevelWord];
+        NSArray *confidences = [helper.tesseract recognizedBlocksByIteratorLevel:G8PageIteratorLevelWord];
         [[[confidences should] have:1] object];
 
         id blockObj = confidences.firstObject;
@@ -128,7 +128,7 @@
     it(@"Should draw blocks on image", ^{
         [helper recognizeImage];
 
-        NSArray *blocks = [helper.tesseract confidencesByIteratorLevel:G8PageIteratorLevelSymbol];
+        NSArray *blocks = [helper.tesseract recognizedBlocksByIteratorLevel:G8PageIteratorLevelSymbol];
         UIImage *blocksImage = [helper.tesseract imageWithBlocks:blocks drawText:YES thresholded:NO];
         UIImage *expectedBlocksImage = [UIImage imageNamed:@"image_sample_bl"];
 
@@ -236,7 +236,7 @@
         [[recognizedText should] containString:kG8WellScanedFirstTitle];
         [[recognizedText shouldNot] containString:kG8WellScanedFinalLongString];
 
-        [[[[helper.tesseract confidencesByIteratorLevel:G8PageIteratorLevelWord] should] haveAtLeast:10] items];
+        [[[[helper.tesseract recognizedBlocksByIteratorLevel:G8PageIteratorLevelWord] should] haveAtLeast:10] items];
     });
 
 });