Improve italic detection for "Binary image compare" OCR - thx tormento :)

+ a few related improvements
See doom9 posts around http://forum.doom9.net/showthread.php?p=1910580#post1910580

Dictionaries/eng_OCRFixReplaceList.xml

@@ -2765,6 +2765,16 @@
     <Word from="AII's" to="All's" />
     <Word from="FIog" to="Flog" />
     <Word from="SIit" to="Slit" />
+    <Word from="allsummer" to="all summer" />
+    <Word from="couldhave" to="could have" />
+    <Word from="flrst" to="first" />
+    <Word from="likeyou" to="like you" />
+    <Word from="makeyou" to="make you" />
+    <Word from="ofmysenioryear" to="of my senior year" />
+    <Word from="thatparty" to="that party" />
+    <Word from="thatpeople" to="that people" />
+    <Word from="thatperson" to="that person" />
+    <Word from="thoughyou" to="though you" />
   </WholeWords>
   <PartialWordsAlways>
     <!-- Will be replaced always -->

LanguageMaster.xml

@@ -2532,7 +2532,7 @@ Keep changes?</KeepChangesMessage>
     <EditImageDb>Edit image db</EditImageDb>
   </VobSubOcr>
   <VobSubOcrCharacter>
-    <Title>VobSub - Manual image to text</Title>
+    <Title>OCR - Manual image to text</Title>
     <ShrinkSelection>Shrink selection</ShrinkSelection>
     <ExpandSelection>Expand selection</ExpandSelection>
     <SubtitleImage>Subtitle image</SubtitleImage>

libse/Language.cs

@@ -2869,7 +2869,7 @@ Keep changes?",
 
             VobSubOcrCharacter = new LanguageStructure.VobSubOcrCharacter
             {
-                Title = "VobSub - Manual image to text",
+                Title = "OCR - Manual image to text",
                 Abort = "&Abort",
                 Skip = "&Skip",
                 SubtitleImage = "Subtitle image",

src/Forms/Ocr/VobSubOcr.cs

@@ -4031,9 +4031,14 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
                     _vobSubOcrCharacter.Initialize(bitmap, item, _manualOcrDialogPosition, _italicCheckedLast, expandSelectionList.Count > 1, null, _lastAdditions, this);
                     DialogResult result = _vobSubOcrCharacter.ShowDialog(this);
                     _manualOcrDialogPosition = _vobSubOcrCharacter.FormPosition;
+                    if (result == DialogResult.Cancel && _vobSubOcrCharacter.SkipImage)
+                    {
+                        break;
+                    }
+                    
                     if (result == DialogResult.OK && _vobSubOcrCharacter.ShrinkSelection)
                     {
-                        shrinkSelection = true;
+                            shrinkSelection = true;
                         index--;
                         if (expandSelectionList.Count > 0)
                         {
@@ -4089,6 +4094,12 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
                         _vobSubOcrCharacter.Initialize(bitmap, item, _manualOcrDialogPosition, _italicCheckedLast, false, bestGuess, _lastAdditions, this, allowExpand);
                         DialogResult result = _vobSubOcrCharacter.ShowDialog(this);
                         _manualOcrDialogPosition = _vobSubOcrCharacter.FormPosition;
+
+                        if (result == DialogResult.Cancel && _vobSubOcrCharacter.SkipImage)
+                        {
+                            break;
+                        }
+
                         if (result == DialogResult.OK && _vobSubOcrCharacter.ExpandSelection)
                         {
                             expandSelectionList.Add(item);
@@ -4171,7 +4182,11 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
                 int wordsNotFound = _ocrFixEngine.CountUnknownWordsViaDictionary(line, out var correctWords);
 
                 // smaller space pixels for italic
-                if (correctWords > 0 && wordsNotFound > 0 && line.Contains("<i>", StringComparison.Ordinal) && matches.Any(p => p?.ImageSplitterItem?.CouldBeSpace == true))
+                if (wordsNotFound > 0 && line.Contains("<i>", StringComparison.Ordinal))
+                {
+                    AddItalicCouldBeSpace(matches, parentBitmap, _unItalicFactor);
+                }
+                if (wordsNotFound > 0 && line.Contains("<i>", StringComparison.Ordinal) && matches.Any(p => p?.ImageSplitterItem?.CouldBeSpace == true))
                 {
                     int j = 0;
                     while (j < matches.Count)
@@ -4196,7 +4211,8 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
                     }
 
                     int tempWordsNotFound = _ocrFixEngine.CountUnknownWordsViaDictionary(tempLine, out var tempCorrectWords);
-                    if (tempWordsNotFound == 0 && tempCorrectWords > 0)
+                    //if (tempWordsNotFound == 0 && tempCorrectWords > 0)
+                    if (tempWordsNotFound < wordsNotFound && tempCorrectWords > 0)
                     {
                         wordsNotFound = tempWordsNotFound;
                         correctWords = tempCorrectWords;
@@ -4253,6 +4269,74 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
             return line;
         }
 
+        private void AddItalicCouldBeSpace(List<CompareMatch> matches, NikseBitmap parentBitmap, double unItalicFactor)
+        {
+            for (int i = 0; i < matches.Count - 1; i++)
+            {
+                var match = matches[i];
+                var matchNext = matches[i + 1];
+                if (!match.Italic || !matchNext.Italic ||
+                    string.IsNullOrWhiteSpace(match.Text) || string.IsNullOrWhiteSpace(matchNext.Text) ||
+                    match.ImageSplitterItem == null || match.ImageSplitterItem.CouldBeSpace)
+                {
+                    continue;
+                }
+
+                int blankVerticalLines = IsVerticalAngledLineTransparent(parentBitmap, match, matchNext, unItalicFactor);
+                if (blankVerticalLines >= _numericUpDownPixelsIsSpace)
+                {
+                    matchNext.ImageSplitterItem.CouldBeSpace = true; // TODO: Rename to "could be space before"
+                }
+            }
+        }
+
+        private int IsVerticalAngledLineTransparent(NikseBitmap parentBitmap, CompareMatch match, CompareMatch next, double unItalicFactor)
+        {
+            int blanks = 0;
+            var min = match.ImageSplitterItem.X + match.ImageSplitterItem.NikseBitmap.Width;
+            var max = next.ImageSplitterItem.X + next.ImageSplitterItem.NikseBitmap.Width / 2;
+            bool abort = false;
+            for (int startX = min; startX < max; startX++)
+            {
+                var lineBlank = true;
+                for (int y = match.ImageSplitterItem.Y; y < match.ImageSplitterItem.Y + match.ImageSplitterItem.NikseBitmap.Height; y++)
+                {
+                    var x = startX - (y - match.ImageSplitterItem.Y) * unItalicFactor;
+                    if (x >= 0)
+                    {
+                        var color = parentBitmap.GetPixel((int)Math.Round(x), y);
+                        if (color.A == 0)
+                        {
+                            // parentBitmap.SetPixel((int)Math.Round(x), y, Color.LawnGreen);
+                        }
+                        else
+                        {
+                            // parentBitmap.SetPixel((int)Math.Round(x), y, Color.Red);
+                            lineBlank = false;
+                            if (blanks > 0)
+                            {
+                                abort = true;
+                                break;
+                            }
+                        }
+                    }
+                }
+
+                if (abort)
+                {
+                    break;
+                }
+
+                if (lineBlank)
+                {
+                    blanks++;
+                }
+            }
+
+            //parentBitmap.GetBitmap().Save(@"J:\Temp\" + DateTime.UtcNow.Ticks + "_" + match.Text + "_" + blanks + ".bmp");
+            return blanks;
+        }
+
         private void SetBinOcrLowercaseUppercase(int height, string text)
         {
             if (text == "e" && (height < _binOcrLastLowercaseHeight || _binOcrLastLowercaseHeight < 0))
@@ -6513,10 +6597,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
                 }
             }
 
-            if (bmp != null)
+            bmp?.Dispose();
-            {
-                bmp.Dispose();
-            }
 
             if (mp.Text != null)
             {
@@ -7003,6 +7084,31 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
                     form.ShowDialog(this);
                 }
             }
+            else if (e.Modifiers == (Keys.Control | Keys.Shift) && e.KeyCode == Keys.I && _ocrMethodIndex == _ocrMethodBinaryImageCompare)
+            {
+                var bmp = (Bitmap)pictureBoxSubtitleImage.Image;
+                if (bmp != null)
+                {
+                    var nBmp = new NikseBitmap(bmp);
+                    bmp.Dispose();
+                    var italicFactor = _unItalicFactor;
+
+                    for (var startX = 20.0; startX < nBmp.Width; startX += 20.0)
+                    {
+                        var x = startX;
+                        for (int y = 0; y < nBmp.Height; y++)
+                        {
+                            x = startX - y * italicFactor;
+                            if (x >= 0)
+                            {
+                                nBmp.SetPixel((int)Math.Round(x), y, Color.Red);
+                            }
+                        }
+                    }
+
+                    pictureBoxSubtitleImage.Image = nBmp.GetBitmap();
+                }
+            }
         }
 
         private void ComboBoxTesseractLanguagesSelectedIndexChanged(object sender, EventArgs e)

src/Forms/Ocr/VobSubOcrCharacter.Designer.cs

@@ -206,6 +206,7 @@
             this.buttonSkip.TabIndex = 3;
             this.buttonSkip.Text = "&Skip";
             this.buttonSkip.UseVisualStyleBackColor = true;
+            this.buttonSkip.Click += new System.EventHandler(this.buttonSkip_Click);
             // 
             // pictureBoxSubtitleImage
             // 

src/Forms/Ocr/VobSubOcrCharacter.cs

@@ -80,10 +80,13 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
 
         public bool ShrinkSelection { get; private set; }
 
+        public bool SkipImage { get; private set; }
+
         internal void Initialize(Bitmap vobSubImage, ImageSplitterItem character, Point position, bool italicChecked, bool showShrink, VobSubOcr.CompareMatch bestGuess, List<VobSubOcr.ImageCompareAddition> additions, VobSubOcr vobSubForm, bool allowExpand = true)
         {
             ShrinkSelection = false;
             ExpandSelection = false;
+            SkipImage = false;
 
             textBoxCharacters.Text = string.Empty;
             if (bestGuess != null)
@@ -292,6 +295,12 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
                 checkBoxItalic.Checked = !checkBoxItalic.Checked;
                 e.SuppressKeyPress = true;
             }
+            else if (e.Modifiers == (Keys.Control | Keys.Shift) && e.KeyCode == Keys.S)
+            {
+                SkipImage = true;
+                DialogResult = DialogResult.Cancel;
+                e.SuppressKeyPress = true;
+            }
         }
 
         private void dataGridView1_CellContentClick(object sender, DataGridViewCellEventArgs e)
@@ -303,5 +312,10 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
         {
             TaskbarList.StopBlink(_vobSubForm);
         }
+
+        private void buttonSkip_Click(object sender, EventArgs e)
+        {
+            DialogResult = DialogResult.Cancel;
+        }
     }
 }

src/Forms/Ocr/VobSubOcrSetItalicFactor.Designer.cs

@@ -34,19 +34,19 @@
             this.pictureBoxSubtitleImage = new System.Windows.Forms.PictureBox();
             this.contextMenuStripListview = new System.Windows.Forms.ContextMenuStrip(this.components);
             this.saveImageAsToolStripMenuItem = new System.Windows.Forms.ToolStripMenuItem();
-            this.numericUpDown1 = new System.Windows.Forms.NumericUpDown();
+            this.numericUpDownItalicFactor = new System.Windows.Forms.NumericUpDown();
             this.labelDescription = new System.Windows.Forms.Label();
             this.saveFileDialog1 = new System.Windows.Forms.SaveFileDialog();
             ((System.ComponentModel.ISupportInitialize)(this.pictureBoxSubtitleImage)).BeginInit();
             this.contextMenuStripListview.SuspendLayout();
-            ((System.ComponentModel.ISupportInitialize)(this.numericUpDown1)).BeginInit();
+            ((System.ComponentModel.ISupportInitialize)(this.numericUpDownItalicFactor)).BeginInit();
             this.SuspendLayout();
             // 
             // buttonCancel
             // 
             this.buttonCancel.Anchor = ((System.Windows.Forms.AnchorStyles)((System.Windows.Forms.AnchorStyles.Bottom | System.Windows.Forms.AnchorStyles.Right)));
             this.buttonCancel.ImeMode = System.Windows.Forms.ImeMode.NoControl;
-            this.buttonCancel.Location = new System.Drawing.Point(680, 312);
+            this.buttonCancel.Location = new System.Drawing.Point(794, 366);
             this.buttonCancel.Name = "buttonCancel";
             this.buttonCancel.Size = new System.Drawing.Size(75, 23);
             this.buttonCancel.TabIndex = 8;
@@ -58,7 +58,7 @@
             // 
             this.buttonOK.Anchor = ((System.Windows.Forms.AnchorStyles)((System.Windows.Forms.AnchorStyles.Bottom | System.Windows.Forms.AnchorStyles.Right)));
             this.buttonOK.ImeMode = System.Windows.Forms.ImeMode.NoControl;
-            this.buttonOK.Location = new System.Drawing.Point(599, 312);
+            this.buttonOK.Location = new System.Drawing.Point(713, 366);
             this.buttonOK.Name = "buttonOK";
             this.buttonOK.Size = new System.Drawing.Size(75, 23);
             this.buttonOK.TabIndex = 7;
@@ -74,7 +74,7 @@
             this.pictureBoxSubtitleImage.ContextMenuStrip = this.contextMenuStripListview;
             this.pictureBoxSubtitleImage.Location = new System.Drawing.Point(12, 47);
             this.pictureBoxSubtitleImage.Name = "pictureBoxSubtitleImage";
-            this.pictureBoxSubtitleImage.Size = new System.Drawing.Size(743, 259);
+            this.pictureBoxSubtitleImage.Size = new System.Drawing.Size(857, 313);
             this.pictureBoxSubtitleImage.SizeMode = System.Windows.Forms.PictureBoxSizeMode.Zoom;
             this.pictureBoxSubtitleImage.TabIndex = 9;
             this.pictureBoxSubtitleImage.TabStop = false;
@@ -93,36 +93,37 @@
             this.saveImageAsToolStripMenuItem.Text = "Save image as...";
             this.saveImageAsToolStripMenuItem.Click += new System.EventHandler(this.saveImageAsToolStripMenuItem_Click);
             // 
-            // numericUpDown1
+            // numericUpDownItalicFactor
             // 
-            this.numericUpDown1.DecimalPlaces = 2;
+            this.numericUpDownItalicFactor.DecimalPlaces = 2;
-            this.numericUpDown1.Increment = new decimal(new int[] {
+            this.numericUpDownItalicFactor.Increment = new decimal(new int[] {
             1,
             0,
             0,
             131072});
-            this.numericUpDown1.Location = new System.Drawing.Point(12, 19);
+            this.numericUpDownItalicFactor.Location = new System.Drawing.Point(12, 19);
-            this.numericUpDown1.Maximum = new decimal(new int[] {
+            this.numericUpDownItalicFactor.Maximum = new decimal(new int[] {
             10,
             0,
             0,
             65536});
-            this.numericUpDown1.Name = "numericUpDown1";
+            this.numericUpDownItalicFactor.Name = "numericUpDownItalicFactor";
-            this.numericUpDown1.Size = new System.Drawing.Size(57, 20);
+            this.numericUpDownItalicFactor.Size = new System.Drawing.Size(57, 20);
-            this.numericUpDown1.TabIndex = 10;
+            this.numericUpDownItalicFactor.TabIndex = 10;
-            this.numericUpDown1.Value = new decimal(new int[] {
+            this.numericUpDownItalicFactor.Value = new decimal(new int[] {
             31,
             0,
             0,
             131072});
-            this.numericUpDown1.ValueChanged += new System.EventHandler(this.numericUpDown1_ValueChanged);
+            this.numericUpDownItalicFactor.ValueChanged += new System.EventHandler(this.numericUpDownItalicFactor_ValueChanged);
             // 
             // labelDescription
             // 
             this.labelDescription.AutoSize = true;
+            this.labelDescription.Font = new System.Drawing.Font("Microsoft Sans Serif", 8.25F, System.Drawing.FontStyle.Bold, System.Drawing.GraphicsUnit.Point, ((byte)(0)));
             this.labelDescription.Location = new System.Drawing.Point(76, 21);
             this.labelDescription.Name = "labelDescription";
-            this.labelDescription.Size = new System.Drawing.Size(431, 13);
+            this.labelDescription.Size = new System.Drawing.Size(529, 13);
             this.labelDescription.TabIndex = 11;
             this.labelDescription.Text = "Adjust value until text style is normal and not italic. Note that original image " +
     "should be italic.";
@@ -131,24 +132,27 @@
             // 
             this.AutoScaleDimensions = new System.Drawing.SizeF(6F, 13F);
             this.AutoScaleMode = System.Windows.Forms.AutoScaleMode.Font;
-            this.ClientSize = new System.Drawing.Size(767, 345);
+            this.BackColor = System.Drawing.Color.DimGray;
+            this.ClientSize = new System.Drawing.Size(881, 399);
             this.Controls.Add(this.labelDescription);
-            this.Controls.Add(this.numericUpDown1);
+            this.Controls.Add(this.numericUpDownItalicFactor);
             this.Controls.Add(this.pictureBoxSubtitleImage);
             this.Controls.Add(this.buttonCancel);
             this.Controls.Add(this.buttonOK);
             this.KeyPreview = true;
             this.MaximizeBox = false;
             this.MinimizeBox = false;
+            this.MinimumSize = new System.Drawing.Size(783, 384);
             this.Name = "VobSubOcrSetItalicFactor";
             this.ShowIcon = false;
             this.ShowInTaskbar = false;
             this.StartPosition = System.Windows.Forms.FormStartPosition.CenterParent;
             this.Text = "Set un-italic factor";
+            this.Shown += new System.EventHandler(this.VobSubOcrSetItalicFactor_Shown);
             this.KeyDown += new System.Windows.Forms.KeyEventHandler(this.VobSubOcrSetItalicFactor_KeyDown);
             ((System.ComponentModel.ISupportInitialize)(this.pictureBoxSubtitleImage)).EndInit();
             this.contextMenuStripListview.ResumeLayout(false);
-            ((System.ComponentModel.ISupportInitialize)(this.numericUpDown1)).EndInit();
+            ((System.ComponentModel.ISupportInitialize)(this.numericUpDownItalicFactor)).EndInit();
             this.ResumeLayout(false);
             this.PerformLayout();
 
@@ -159,7 +163,7 @@
         private System.Windows.Forms.Button buttonCancel;
         private System.Windows.Forms.Button buttonOK;
         private System.Windows.Forms.PictureBox pictureBoxSubtitleImage;
-        private System.Windows.Forms.NumericUpDown numericUpDown1;
+        private System.Windows.Forms.NumericUpDown numericUpDownItalicFactor;
         private System.Windows.Forms.Label labelDescription;
         private System.Windows.Forms.SaveFileDialog saveFileDialog1;
         private System.Windows.Forms.ContextMenuStrip contextMenuStripListview;

src/Forms/Ocr/VobSubOcrSetItalicFactor.cs

@@ -19,7 +19,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
 
             _bmp = bmp;
             _factor = factor;
-            numericUpDown1.Value = (decimal)factor;
+            numericUpDownItalicFactor.Value = (decimal)factor;
 
             Text = Configuration.Settings.Language.VobSubOcrSetItalicFactor.Title;
             labelDescription.Text = Configuration.Settings.Language.VobSubOcrSetItalicFactor.Description;
@@ -28,9 +28,9 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
             buttonCancel.Text = Configuration.Settings.Language.General.Cancel;
         }
 
-        private void numericUpDown1_ValueChanged(object sender, EventArgs e)
+        private void numericUpDownItalicFactor_ValueChanged(object sender, EventArgs e)
         {
-            pictureBoxSubtitleImage.Image = VobSubOcr.UnItalic(_bmp, (double)numericUpDown1.Value);
+            pictureBoxSubtitleImage.Image = VobSubOcr.UnItalic(_bmp, (double)numericUpDownItalicFactor.Value);
         }
 
         internal double GetUnItalicFactor()
@@ -40,7 +40,7 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
 
         private void buttonOK_Click(object sender, EventArgs e)
         {
-            _factor = (double)numericUpDown1.Value;
+            _factor = (double)numericUpDownItalicFactor.Value;
             DialogResult = DialogResult.OK;
         }
 
@@ -101,5 +101,9 @@ namespace Nikse.SubtitleEdit.Forms.Ocr
             }
         }
 
+        private void VobSubOcrSetItalicFactor_Shown(object sender, EventArgs e)
+        {
+            numericUpDownItalicFactor_ValueChanged(null, null);
+        }
     }
 }