Improve pixel density logic in OcrdExif, OCR-D#256, OCR-D#37, OCR-D/o…

…crd_tesserocr#54
bertsky · Jul 11, 2019 · ef37dc4 · ef37dc4
1 parent 6f925e3
commit ef37dc4
Show file tree

Hide file tree

Showing 3 changed files with 27 additions and 3 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -5,6 +5,10 @@ Versioned according to [Semantic Versioning](http://semver.org/).
 
 ## Unreleased
 
+Changed:
+
+  * Improve pixel density logic in OcrdExif, #256, #37, OCR-D/ocrd_tesserocr#54
+
 ## [1.0.0b10] - 2019-06-25
 
 Fixed:

diff --git a/ocrd_models/ocrd_models/ocrd_exif.py b/ocrd_models/ocrd_models/ocrd_exif.py
@@ -2,9 +2,22 @@
 Technical image metadata
 """
 
+from math import sqrt
+
 class OcrdExif():
-    """
-    Represents technical image metadata
+    """Represents technical image metadata.
+    
+    Members:
+    - `width` / `height`: pixel dimensions
+    - `photometricInterpretation`: pixel type/depth, e.g.
+      '1' for b/w,
+      'L' for 8-bit grayscale,
+      'RGB' for 24-bit truecolor,
+      'I' for 32-bit signed integer grayscale,
+      'F' for floating-point grayscale
+      (see PIL concept `mode`)
+    - `resolution` / `xResolution` / `yResolution`: pixel density
+    - `resolutionUnits`: unit of measurement (either `inches` or `cm`)
     """
 
     def __init__(self, img):
@@ -27,7 +40,7 @@ def __init__(self, img):
         elif img.format == 'JPEG':
             self.xResolution = img.info['jfif_density'][0]
             self.yResolution = img.info['jfif_density'][1]
-            self.resolutionUnit = img.info['jfif_unit']
+            self.resolutionUnit = 'cm' if img.info['jfif_unit'] == 2 else 'inches'
         elif img.format == 'PNG' and 'aspect' in img.info:
             self.xResolution = img.info['aspect'][0]
             self.yResolution = img.info['aspect'][1]
@@ -38,6 +51,7 @@ def __init__(self, img):
             self.xResolution = 1
             self.yResolution = 1
             self.resolutionUnit = 'inches'
+        self.resolution = round(sqrt(self.xResolution * self.yResolution))
 
     def to_xml(self):
         """

diff --git a/tests/model/test_exif.py b/tests/model/test_exif.py
@@ -22,6 +22,7 @@ def test_tiff(self):
         self.assertEqual(exif.height, 3749)
         self.assertEqual(exif.xResolution, 300)
         self.assertEqual(exif.yResolution, 300)
+        self.assertEqual(exif.resolution, 300)
         self.assertEqual(exif.compression, 'jpeg')
         self.assertEqual(exif.photometricInterpretation, 'RGB')
 
@@ -31,6 +32,7 @@ def test_png1(self):
         self.assertEqual(exif.height, 2084)
         self.assertEqual(exif.xResolution, 300)
         self.assertEqual(exif.yResolution, 300)
+        self.assertEqual(exif.resolution, 300)
         self.assertEqual(exif.compression, None)
         self.assertEqual(exif.photometricInterpretation, '1')
 
@@ -40,6 +42,7 @@ def test_png2(self):
         self.assertEqual(exif.height, 3062)
         self.assertEqual(exif.xResolution, 1)
         self.assertEqual(exif.yResolution, 1)
+        self.assertEqual(exif.resolution, 1)
         self.assertEqual(exif.photometricInterpretation, '1')
 
     def test_jpg(self):
@@ -48,6 +51,8 @@ def test_jpg(self):
         self.assertEqual(exif.height, 1472)
         self.assertEqual(exif.xResolution, 1)
         self.assertEqual(exif.yResolution, 1)
+        self.assertEqual(exif.resolution, 1)
+        self.assertEqual(exif.resolutionUnit, 'inches')
         self.assertEqual(exif.photometricInterpretation, 'RGB')
 
     def test_jp2(self):
@@ -56,6 +61,7 @@ def test_jp2(self):
         self.assertEqual(exif.height, 2084)
         self.assertEqual(exif.xResolution, 1)
         self.assertEqual(exif.yResolution, 1)
+        self.assertEqual(exif.resolution, 1)
         self.assertEqual(exif.photometricInterpretation, 'RGB')
 
 if __name__ == '__main__':