[WIP] Add deskewing per tesseract

With the help of `OSD`, tesseract determines the skew angle for
images. The wrapper applies this to pages and regions. It is
not clear yet how to save the estimated skew angle in PAGE
XML. Cf. PRImA-Research-Lab/PAGE-XML#9

ocrd_tesserocr/__init__.py

@@ -1,3 +1,4 @@
 from .recognize import TesserocrRecognize
 from .segment_line import TesserocrSegmentLine
 from .segment_region import TesserocrSegmentRegion
+from .deskew import TesserocrDeskew

ocrd_tesserocr/cli.py

@@ -5,6 +5,7 @@
 from ocrd_tesserocr.segment_region import TesserocrSegmentRegion
 from ocrd_tesserocr.segment_line import TesserocrSegmentLine
 from ocrd_tesserocr.segment_word import TesserocrSegmentWord
+from ocrd_tesserocr.deskew import TesserocrDeskew
 
 @click.command()
 @ocrd_cli_options
@@ -25,3 +26,8 @@ def ocrd_tesserocr_segment_word(*args, **kwargs):
 @ocrd_cli_options
 def ocrd_tesserocr_recognize(*args, **kwargs):
     return ocrd_cli_wrap_processor(TesserocrRecognize, *args, **kwargs)
+
+@click.command()
+@ocrd_cli_options
+def ocrd_tesserocr_deskew(*args, **kwargs):
+    return ocrd_cli_wrap_processor(TesserocrDeskew, *args, **kwargs)

ocrd_tesserocr/deskew.py

@@ -0,0 +1,91 @@
+from __future__ import absolute_import
+
+import locale
+
+# pylint: disable=wrong-import-position
+locale.setlocale(locale.LC_ALL, 'C') # circumvent tesseract-ocr issue 1670 (which cannot be done on command line because Click requires an UTF-8 locale in Python 3)
+
+from tesserocr import RIL, PSM, PyTessBaseAPI
+
+from ocrd_utils import getLogger, concat_padded, xywh_from_points, points_from_x0y0x1y1, MIMETYPE_PAGE
+from ocrd_modelfactory import page_from_file
+from ocrd_models.ocrd_page import (
+    CoordsType,
+    GlyphType,
+    LabelType,
+    LabelsType,
+    MetadataItemType,
+    TextEquivType,
+    TextStyleType,
+
+    to_xml
+)
+from ocrd import Processor
+from .config import TESSDATA_PREFIX, OCRD_TOOL
+
+log = getLogger('processor.TesserocrDeskew')
+
+class TesserocrDeskew(Processor):
+
+    def __init__(self, *args, **kwargs):
+        kwargs['ocrd_tool'] = OCRD_TOOL['tools']['ocrd-tesserocr-deskew']
+        kwargs['version'] = OCRD_TOOL['version']
+        super(TesserocrDeskew, self).__init__(*args, **kwargs)
+
+    def process(self):
+        """
+        Performs the deskewing.
+        """
+        # print(self.parameter)
+        oplevel = self.parameter['operation_level']
+        with PyTessBaseAPI(path=TESSDATA_PREFIX, psm=PSM.AUTO_OSD) as tessapi:
+            for (n, input_file) in enumerate(self.input_files):
+                log.info("INPUT FILE %i / %s", n, input_file)
+                pcgts = page_from_file(self.workspace.download_file(input_file))
+                pil_image = self.workspace.resolve_image_as_pil(pcgts.get_Page().imageFilename)
+
+                metadata = pcgts.get_Metadata() # ensured by from_file()
+                metadata.add_MetadataItem(
+                    MetadataItemType(type_="processingStep",
+                                     name=OCRD_TOOL['tools']['ocrd-tesserocr-deskew']['steps'][0],
+                                     value='ocrd-tesserocr-deskew',
+                                     Labels=[LabelsType(externalRef="parameters",
+                                                        Label=[LabelType(type_=name,
+                                                                         value=self.parameter[name])
+                                                               for name in self.parameter.keys()])]))
+                log.info("Deskewing on '%s' level on page '%s'", oplevel, pcgts.get_pcGtsId())
+
+                if oplevel == 'page':
+                    self._process_page(tessapi, pil_image)
+                elif oplevel == 'region': 
+                    regions = pcgts.get_Page().get_TextRegion()
+                    if not regions:
+                        log.warning("Deskewing regions requested but page contains no text regions")
+                    self._process_regions(regions, tessapi, pil_image)
+
+                ID = concat_padded(self.output_file_grp, n)
+                self.workspace.add_file(
+                    ID=ID,
+                    file_grp=self.output_file_grp,
+                    mimetype=MIMETYPE_PAGE,
+                    local_filename='%s/%s' % (self.output_file_grp, ID),
+                    content=to_xml(pcgts),
+                )
+
+    def _process_page(self, tessapi, pil_image):
+        tessapi.SetImage(pil_image)
+        orientation, direction, order, deskew_angle = tessapi.AnalyseLayout().Orientation()
+        log.debug("Deskew angle: {:.4f}".format(deskew_angle))
+
+    def _process_regions(self, regions, tessapi, pil_image):
+        for region in regions:
+            log.debug("Deskewing region '%s'", region.id)
+            region_xywh = xywh_from_points(region.get_Coords().points)
+
+            # Note: we set the image instead of specifying a rectangle!
+            pil_region_image = pil_image.crop((region_xywh['x'], region_xywh['y'], region_xywh['x'] + region_xywh['w'], region_xywh['y'] + region_xywh['h']))
+            tessapi.SetImage(pil_region_image)
+
+            orientation, direction, order, deskew_angle = tessapi.AnalyseLayout().Orientation()
+            log.debug("Deskew angle: {:.4f}".format(deskew_angle))
+            region.set_orientation(deskew_angle)

ocrd_tesserocr/ocrd-tool.json

@@ -3,6 +3,28 @@
   "git_url": "https://github.com/OCR-D/ocrd_tesserocr",
   "dockerhub": "ocrd/tesserocr",
   "tools": {
+    "ocrd-tesserocr-deskew": {
+      "executable": "ocrd-tesserocr-deskew",
+      "categories": ["Deskewing"],
+      "description": "Deskew images or image parts",
+      "input_file_grp": [
+        "OCR-D-IMG",
+        "OCR-D-SEG-BLOCK"
+      ],
+      "output_file_grp": [
+        "OCR-D-DESKEW-IMG",
+        "OCR-D-DESKEW-BLOCK"
+      ],
+      "steps": ["preprocessing/optimization/deskewing"],
+      "parameters": {
+        "operation_level": {
+          "type": "string",
+          "enum": ["page","region"],
+          "default": "page",
+          "description": "Level of operation for deskewing"
+        }
+      }
+    },
     "ocrd-tesserocr-recognize": {
       "executable": "ocrd-tesserocr-recognize",
       "categories": ["Text recognition and optimization"],

setup.py

@@ -1,11 +1,12 @@
 # -*- coding: utf-8 -*-
 """
-Installs four executables:
+Installs five executables:
 
     - ocrd_tesserocr_recognize
     - ocrd_tesserocr_segment_region
     - ocrd_tesserocr_segment_line
     - ocrd_tesserocr_segment_word
+    - ocrd_tesserocr_deskew
 """
 import codecs
 
@@ -16,9 +17,9 @@
     version='0.2.1',
     description='Tesserocr bindings',
     long_description=codecs.open('README.rst', encoding='utf-8').read(),
-    author='Konstantin Baierer',
-    author_email='[email protected]',
-    url='https://github.com/kba/ocrd_tesserocr',
+    author='Konstantin Baierer, Kay-Michael Würzner',
+    author_email='[email protected], [email protected]',
+    url='https://github.com/OCR-D/ocrd_tesserocr',
     license='Apache License 2.0',
     packages=find_packages(exclude=('tests', 'docs')),
     install_requires=open('requirements.txt').read().split('\n'),
@@ -31,6 +32,7 @@
             'ocrd-tesserocr-segment-region=ocrd_tesserocr.cli:ocrd_tesserocr_segment_region',
             'ocrd-tesserocr-segment-line=ocrd_tesserocr.cli:ocrd_tesserocr_segment_line',
             'ocrd-tesserocr-segment-word=ocrd_tesserocr.cli:ocrd_tesserocr_segment_word',
+            'ocrd-tesserocr-deskew=ocrd_tesserocr.cli:ocrd_tesserocr_deskew',
         ]
     },
 )