private void _pagesListBox_SelectedIndexChanged(object sender, EventArgs e) { // Get the recognized words of the selected page _wordsListBox.Items.Clear(); IOcrPage ocrPage = _ocrDocument.Pages[_pagesListBox.SelectedIndex]; if (ocrPage.IsRecognized) { IOcrPageCharacters pageCharacters = ocrPage.GetRecognizedCharacters(); if (pageCharacters == null) { return; } foreach (IOcrZoneCharacters zoneCharacters in pageCharacters) { ICollection <OcrWord> words = zoneCharacters.GetWords(); foreach (OcrWord word in words) { _wordsListBox.Items.Add(word.Value); } } } }
private static double GetConfidence(IOcrPage page) { var recognizedCharacters = page.GetRecognizedCharacters(); var findZoneCharacters = recognizedCharacters.FindZoneCharacters(0); return(findZoneCharacters .DefaultIfEmpty() .Average(character => character.Confidence) / 100D); }
private void BuildWordLists() { _ocrPageCharacters = _ocrPage.GetRecognizedCharacters(); _ocrZoneWords = new List <List <OcrWord> >(); // Build the words foreach (IOcrZoneCharacters zoneCharacters in _ocrPageCharacters) { List <OcrWord> words = new List <OcrWord>(); words.AddRange(zoneCharacters.GetWords()); _ocrZoneWords.Add(words); } _selectedZoneIndex = -1; _selectedWordIndex = -1; }
private void GetOmrReading(IOcrPage ocrPage, FormField field, ImageField imageField, int retry = 1) { IOcrPageCharacters pageCharacters = ocrPage.GetRecognizedCharacters(); if (pageCharacters == null) { logger.Warn($"could not read OMR for ${field} "); imageField.FieldResult.Confidence = 0; imageField.FieldResult.Text = ""; } else { IOcrZoneCharacters zoneCharacters = pageCharacters[0]; if (zoneCharacters.Count > 0) { OcrCharacter omrCharacter = zoneCharacters[0]; imageField.FieldResult.Text = omrCharacter.Code.ToString(); imageField.FieldResult.IsFilled = omrCharacter.Code == FilledChar; imageField.FieldResult.Confidence = omrCharacter.Confidence; // often on a fill we get the line from the box, so we retry more narrowly if (imageField.FieldResult.IsFilled) { if (retry > 0) { var orgZone = ocrPage.Zones[0]; orgZone.Bounds = ChangeBoundsRatio(orgZone.Bounds, 0.66); ocrPage.Recognize(null); GetOmrReading(ocrPage, field, imageField, 0); logger.Info($"FILLED {field.Name}"); } } } else { imageField.FieldResult.Text = ""; } } }
public void ZoneToControls(int index) { // Fill the controls from the current zone if (index != -1) { OcrZone zone = _zones[index]; _nameTextBox.Text = zone.Name; // Convert the bounds to pixels LeadRect bounds = zone.Bounds; _leftTextBox.Text = bounds.X.ToString(); _topTextBox.Text = bounds.Y.ToString(); _widthTextBox.Text = bounds.Width.ToString(); _heightTextBox.Text = bounds.Height.ToString(); // Disable these events when changing the combo boxes selected items once the "UpdateZonesControl" gets activated this._typeComboBox.SelectedIndexChanged -= new System.EventHandler(this._propertiesComboBox_SelectedIndexChanged); this._languageComboBox.SelectedIndexChanged -= new System.EventHandler(this._propertiesComboBox_SelectedIndexChanged); this._zoneViewPerspectiveComboBox.SelectedIndexChanged -= new System.EventHandler(this._propertiesComboBox_SelectedIndexChanged); this._zoneTextDirectionComboBox.SelectedIndexChanged -= new System.EventHandler(this._propertiesComboBox_SelectedIndexChanged); _typeComboBox.SelectedItem = zone.ZoneType; for (int i = 0; i < _languageComboBox.Items.Count; i++) { MyLanguage ml = (MyLanguage)_languageComboBox.Items[i]; if (zone.Language == null || zone.Language == String.Empty) { if (ml.Language == String.Empty) { _languageComboBox.SelectedItem = ml; break; } } else { if (ml.Language == zone.Language) { _languageComboBox.SelectedItem = ml; break; } } } _zoneViewPerspectiveComboBox.SelectedIndex = 0; foreach (ViewPerspectiveItem item in _zoneViewPerspectiveComboBox.Items) { if (item.ViewPerspective == zone.ViewPerspective) { _zoneViewPerspectiveComboBox.SelectedItem = item; break; } } _zoneTextDirectionComboBox.SelectedIndex = 0; foreach (TextDirectionItem item in _zoneTextDirectionComboBox.Items) { if (item.TextDirection == zone.TextDirection) { _zoneTextDirectionComboBox.SelectedItem = item; break; } } this._typeComboBox.SelectedIndexChanged += new System.EventHandler(this._propertiesComboBox_SelectedIndexChanged); this._languageComboBox.SelectedIndexChanged += new System.EventHandler(this._propertiesComboBox_SelectedIndexChanged); this._zoneViewPerspectiveComboBox.SelectedIndexChanged += new System.EventHandler(this._propertiesComboBox_SelectedIndexChanged); this._zoneTextDirectionComboBox.SelectedIndexChanged += new System.EventHandler(this._propertiesComboBox_SelectedIndexChanged); if (zone.ZoneType == OcrZoneType.Omr) { StringBuilder sb = new StringBuilder(); if (!_ocrPage.IsRecognized) { sb.Append("Unfilled (0% certain)"); } else { IOcrPageCharacters pageCharacters = _ocrPage.GetRecognizedCharacters(); if (pageCharacters == null || pageCharacters.Count == 0 || zone.Id >= pageCharacters.Count) { sb.Append("Unfilled (0% certain)"); } else { IOcrZoneCharacters zoneCharacters = pageCharacters[zone.Id]; if (zoneCharacters.Count > 0) { OcrCharacter omrCharacter = zoneCharacters[0]; char filledChar = _ocrEngine.ZoneManager.OmrOptions.GetStateRecognitionCharacter(OcrOmrZoneState.Filled); char unfilledChar = _ocrEngine.ZoneManager.OmrOptions.GetStateRecognitionCharacter(OcrOmrZoneState.Unfilled); if (omrCharacter.Code == filledChar) { sb.Append("Filled"); } else { sb.Append("Unfilled"); } sb.AppendFormat(" ({0}% certain)", omrCharacter.Confidence); } else { sb.AppendFormat("Unfilled (0% certain)"); } } } _omrStatusLabel.Text = sb.ToString(); } else { _omrStatusLabel.Text = string.Empty; } if ((zone.CharacterFilters & OcrZoneCharacterFilters.Digit) == OcrZoneCharacterFilters.Digit) { _digitCheckBox.Checked = true; } else { _digitCheckBox.Checked = false; } if ((zone.CharacterFilters & OcrZoneCharacterFilters.Plus) == OcrZoneCharacterFilters.Plus) { _plusCheckBox.Checked = true; } else { _plusCheckBox.Checked = false; } } else { _nameTextBox.Text = string.Empty; _leftTextBox.Text = string.Empty; _topTextBox.Text = string.Empty; _widthTextBox.Text = string.Empty; _heightTextBox.Text = string.Empty; _typeComboBox.SelectedIndex = 0; _languageComboBox.SelectedIndex = 0; _zoneViewPerspectiveComboBox.SelectedIndex = 0; _zoneTextDirectionComboBox.SelectedIndex = 0; _omrStatusLabel.Text = string.Empty; _digitCheckBox.Checked = false; _plusCheckBox.Checked = false; } }
private static double GetConfidence(IOcrPage page) { var recognizedCharacters = page.GetRecognizedCharacters(); var findZoneCharacters = recognizedCharacters.FindZoneCharacters(0); return findZoneCharacters .DefaultIfEmpty() .Average(character => character.Confidence) / 100D; }
private PageResults GetPageConfidence(IOcrPage ocrPage) { IOcrPageCharacters pageCharacters = ocrPage.GetRecognizedCharacters(); double pageConfidence = 0; int certainWords = 0; int totalWords = 0; int totalZoneWords = 0; int textZoneCount = 0; for (int i = 0; i < ocrPage.Zones.Count; i++) { IOcrZoneCharacters zoneCharacters = pageCharacters.FindZoneCharacters(i); if (zoneCharacters.Count == 0) { continue; } textZoneCount++; double zoneConfidence = 0; int characterCount = 0; double wordConfidence = 0; totalZoneWords = 0; bool newWord = true; foreach (var ocrCharacter in zoneCharacters) { if (newWord) { wordConfidence = 0; characterCount = 0; wordConfidence = 1000; } if (ocrCharacter.Confidence < wordConfidence) { wordConfidence = ocrCharacter.Confidence; } characterCount++; if ((ocrCharacter.Position & OcrCharacterPosition.EndOfWord) == OcrCharacterPosition.EndOfWord || (ocrCharacter.Position & OcrCharacterPosition.EndOfLine) == OcrCharacterPosition.EndOfLine) { if (characterCount > 3) { if (ocrCharacter.WordIsCertain) { certainWords++; } totalWords++; totalZoneWords++; zoneConfidence += wordConfidence; } newWord = true; } else { newWord = false; } } if (totalZoneWords > 0) { zoneConfidence /= totalZoneWords; pageConfidence += zoneConfidence; } else { zoneConfidence = 0; pageConfidence += zoneConfidence; } } if (textZoneCount > 0) { pageConfidence /= textZoneCount; } else { pageConfidence = 0; } PageResults results = new PageResults(pageConfidence, certainWords, totalWords); return(results); }