/// <summary>
/// return true if the oContainedRect contained this
/// </summary>
/// <param name="oContainedRect">The o contained rect.</param>
/// <returns></returns>
public Boolean IsInside(TOCRRect oContainedRect)
{
return((m_iLeft + s_iXThreshold >= oContainedRect.Left) &&
(Right - s_iXThreshold <= oContainedRect.Right) &&
(m_iTop + s_iYThreshold >= oContainedRect.Top) &&
(Bottom - s_iYThreshold <= oContainedRect.Bottom));
}
// Check 2 rect overlapping & return the overlapping data
// oRect - (input) The rect to check
// oOverlapRect - (output) The overlapping area
/// <summary>
/// Overlaps the rect.
/// </summary>
/// <param name="oRect">The o rect.</param>
/// <param name="oOverlapRect">The o overlap rect.</param>
/// <returns></returns>
public Boolean OverlapRect(TOCRRect oRect,
TOCRRect oOverlapRect)
{
if (!IsOverlap(oRect))
{
return(false);
}
int iNewLeft = Math.Max(m_iLeft, oRect.m_iLeft);
int iNewTop = Math.Max(m_iTop, oRect.m_iTop);
oOverlapRect.m_iWidth = Math.Min(Right, oRect.Right) - iNewLeft;
oOverlapRect.m_iHeight = Math.Min(Bottom, oRect.Bottom) - iNewTop;
oOverlapRect.m_iLeft = iNewLeft;
oOverlapRect.m_iTop = iNewTop;
// In case there is intersect only with the help of the thresholds the we create "small" rectangle in the intersect area
if (oOverlapRect.m_iWidth <= 0)
{
oOverlapRect.m_iWidth = 1;
}
if (oOverlapRect.m_iHeight <= 0)
{
oOverlapRect.m_iHeight = 1;
}
return(true);
}
// Add a rectangle (Union rectangles)
/// <summary>
/// Adds the specified o rect.
/// </summary>
/// <param name="oRect">The o rect.</param>
public void Add(TOCRRect oRect)
{
// Do nothing if the add rect is empty
if (oRect.IsEmpty())
{
return;
}
// Copy the add rect of the target rect is empty
if (IsEmpty())
{
m_iLeft = oRect.m_iLeft;
m_iTop = oRect.m_iTop;
m_iWidth = oRect.m_iWidth;
m_iHeight = oRect.m_iHeight;
}
else
{
// Union rectangles...
int iNewRight = Math.Max(Right, oRect.Right);
int iNewBottom = Math.Max(Bottom, oRect.Bottom);
m_iLeft = Math.Min(m_iLeft, oRect.m_iLeft);
m_iTop = Math.Min(m_iTop, oRect.m_iTop);
m_iWidth = iNewRight - m_iLeft;
m_iHeight = iNewBottom - m_iTop;
}
}
//
/// <summary>
/// Determines whether the specified and other word are related.
/// </summary>
/// <param name="oOtherWord">The other word.</param>
/// <returns>
/// True if other word is "related" to the current word
/// </returns>
public Boolean IsRelated(TWord oOtherWord)
{
// First check we do not have ant chance for intersection
if (!m_oRectangle.IsOverlap(oOtherWord.m_oRectangle))
{
return(false);
}
foreach (TChar oChar in this)
{
foreach (TChar oOtherChar in oOtherWord)
{
TOCRRect oOverlapRect = new TOCRRect();
if (oChar.Rect.OverlapRect(oOtherChar.Rect, oOverlapRect))
{
if (oOverlapRect.Area * 2 > oChar.Rect.Area)
{
return(true);
}
}
}
}
return(false);
}
/// <summary>
/// Initializes a new instance of the <see cref="TOCRRect"/> class.
/// </summary>
/// <param name="oRect">The o rect.</param>
public TOCRRect(TOCRRect oRect)
{
m_iLeft = oRect.m_iLeft;
m_iTop = oRect.m_iTop;
m_iWidth = oRect.m_iWidth;
m_iHeight = oRect.m_iHeight;
}
/// <summary>
/// Rotate the rectangle to be horizontal
/// </summary>
/// <param name="oRect">The o rect.</param>
/// <returns></returns>
public TOCRRect Deskew(TOCRRect oRect)
{
return(new TOCRRect(
oRect.Left + (oRect.Top * m_iDeskewY) / m_iDeskewX,
oRect.Top - (oRect.Left * m_iDeskewY) / m_iDeskewX,
oRect.Width,
oRect.Height));
}
//
// Public methods
//
/// <summary>
/// Initializes a new instance of the <see cref="TWord" /> class.
/// </summary>
public TWord()
{
m_oChars = new TChars();
m_oRectangle = new TOCRRect();
m_sWordData = "";
m_iConfidance = 0;
m_iStyle = 0;
}
//
// Public methods
//
/// <summary>
/// Initializes a new instance of the <see cref="TPage"/> class.
/// </summary>
public TPage()
{
m_oLines = new List <TLine>();
m_oRect = new TOCRRect();
// Default values...
m_iDeskewY = 0;
m_iDeskewX = 4000;
m_iResolution = 1;
}
/// <summary>
/// Initializes a new instance of the <see cref="TChar"/> class.
/// </summary>
/// <param name="cCharData">The c character data.</param>
/// <param name="iConfidance">The i confidance.</param>
/// <param name="oRectangle">The o rectangle.</param>
public TChar(
char cCharData,
short iConfidance,
TOCRRect oRectangle)
: this()
{
m_oRectangle = new TOCRRect(oRectangle);
m_oCharCandidate.Add(new TBasicChar(cCharData, iConfidance));
}
/// <summary>
/// Determines whether the specified o other rect is equal.
/// </summary>
/// <param name="oOtherRect">The o other rect.</param>
/// <returns></returns>
public Boolean IsEqual(TOCRRect oOtherRect)
{
int thresholdX = Math.Min(s_iXThreshold, Width / 2);
int thresholdY = Math.Min(s_iYThreshold, Height / 2);
return((Math.Abs(m_iLeft - oOtherRect.m_iLeft) <= thresholdX) &&
(Math.Abs(Right - oOtherRect.Right) <= thresholdX) &&
(Math.Abs(m_iTop - oOtherRect.m_iTop) <= thresholdY) &&
(Math.Abs(Bottom - oOtherRect.Bottom) <= thresholdY));
}
/// <summary>
/// Initializes a new instance of the <see cref="TChar"/> class.
/// </summary>
/// <param name="cCharData">The c character data.</param>
/// <param name="iConfidance">The i confidance.</param>
/// <param name="cCharDataCand2">The c character data cand2.</param>
/// <param name="iConfidance2">The i confidance2.</param>
/// <param name="oRectangle">The o rectangle.</param>
public TChar(
char cCharData,
short iConfidance,
char cCharDataCand2,
short iConfidance2,
TOCRRect oRectangle)
: this(cCharData, iConfidance, oRectangle)
{
m_oCharCandidate.Add(new TBasicChar(cCharDataCand2, iConfidance2));
}
// return true if the 2 rect have overlapping area (including the threshold )
/// <summary>
/// Determines whether the specified o rect is overlap.
/// </summary>
/// <param name="oRect">The o rect.</param>
/// <returns></returns>
public Boolean IsOverlap(TOCRRect oRect)
{
if ((m_iLeft - s_iXThreshold > oRect.Right) ||
(Right + s_iXThreshold < oRect.m_iLeft) ||
(m_iTop - s_iYThreshold > oRect.Bottom) ||
(Bottom + s_iYThreshold < oRect.m_iTop))
{
return(false);
}
else
{
return(true);
}
}
/// <summary>
/// Recieves BinaryReader and page, and initialize the page from the reader.
/// </summary>
public void ReadData(System.IO.BinaryReader binaryReader)
{
m_iDeskewY = binaryReader.ReadInt32();
m_iDeskewX = binaryReader.ReadInt32();
m_iResolution = binaryReader.ReadInt32();
m_iImageWidth = binaryReader.ReadInt32();
m_iImageHeight = binaryReader.ReadInt32();
// Read Rectangle data
int rectangleTop = binaryReader.ReadInt32();
int rectangleHeight = binaryReader.ReadInt32();
int rectangleLeft = binaryReader.ReadInt32();
int rectangleWidth = binaryReader.ReadInt32();
m_oRect = new TOCRRect(rectangleLeft, rectangleTop, rectangleWidth, rectangleHeight);
}
public int Compare(TLine x, TLine y)
{
TLine oLine1 = x;
TLine oLine2 = y;
if ((oLine1.Words.Count > 0) && (oLine2.Words.Count > 0))
{
TOCRRect oRect1 = m_oPage.Deskew(oLine1.Word(0).Rect);
TOCRRect oRect2 = m_oPage.Deskew(oLine2.Word(0).Rect);
return(oRect1.Top - oRect2.Top);
}
else
{
TOCRRect oRect1 = m_oPage.Deskew(oLine1.Rect);
TOCRRect oRect2 = m_oPage.Deskew(oLine2.Rect);
return(oRect1.Top - oRect2.Top);
}
}
// Add candidate (remove duplicated is exist)
/// <summary>
/// Adds the candidate.
/// </summary>
/// <param name="oNewCandidate">The o new candidate.</param>
public void AddCandidate(TChar oNewCandidate)
{
foreach (TBasicChar oNewChar in oNewCandidate)
{
Boolean bAdd = false;
int iIdx = -1;
// Check if we already have candidate with the same data
for (int i = 0; (i < m_oCharCandidate.Count) && (iIdx < 0); i++)
{
TBasicChar oChar = (TBasicChar)m_oCharCandidate[i];
if (oChar.CharData == oNewChar.CharData)
{
iIdx = i;
}
}
//int iIdx = m_oCharCandidate.BinarySearch( oNewChar ) ; xxx
if (iIdx >= 0)
{
TBasicChar oExistChar = (TBasicChar)(m_oCharCandidate[iIdx]);
// The same character have lower confidence - so we now replace the old with the new
if (oExistChar.CompareTo(oNewChar) > 0)
{
m_oCharCandidate.RemoveAt(iIdx);
bAdd = true;
}
}
else
{
bAdd = true;
}
if (bAdd)
{
// Add the new candidate (& sort acurding the candidates confidence )
m_oCharCandidate.Add(oNewChar);
m_oCharCandidate.Sort();
// In case the new candidate is the best, then we take also the char rectangle.
if (m_oCharCandidate[0] == oNewChar)
{
m_oRectangle = oNewCandidate.m_oRectangle;
}
}
}
}
//
// Public methods
//
/// <summary>
/// Initializes a new instance of the <see cref="TChar"/> class.
/// </summary>
public TChar()
{
m_oRectangle = new TOCRRect();
m_oCharCandidate = new List <TBasicChar>();
}
/// <summary>
/// Determines whether [is equal bottom] [the specified o other rect].
/// </summary>
/// <param name="oOtherRect">The o other rect.</param>
/// <returns></returns>
public Boolean IsEqualBottom(TOCRRect oOtherRect)
{
return(Math.Abs(Bottom - oOtherRect.Bottom) <= s_iYThreshold);
}
/// <summary>
/// Determines whether [is equal left] [the specified o other rect].
/// </summary>
/// <param name="oOtherRect">The o other rect.</param>
/// <returns></returns>
public Boolean IsEqualLeft(TOCRRect oOtherRect)
{
return(Math.Abs(m_iLeft - oOtherRect.m_iLeft) <= s_iXThreshold);
}
/// <summary>
/// Determines whether [is equal right] [the specified o other rect].
/// </summary>
/// <param name="oOtherRect">The o other rect.</param>
/// <returns></returns>
public Boolean IsEqualRight(TOCRRect oOtherRect)
{
return(Math.Abs(Right - oOtherRect.Right) <= s_iXThreshold);
}
/// <summary>
/// Determines whether [is equal top] [the specified o other rect].
/// </summary>
/// <param name="oOtherRect">The o other rect.</param>
/// <returns></returns>
public Boolean IsEqualTop(TOCRRect oOtherRect)
{
return(Math.Abs(m_iTop - oOtherRect.m_iTop) <= s_iYThreshold);
}
/// <summary>
/// Initializes a new instance of the <see cref="TLine"/> class.
/// </summary>
public TLine()
{
m_oWords = new System.Collections.Generic.List <TWord>();
m_oRect = new TOCRRect();
m_oComparerWordLeft = new TWord.TComparerLeft();
}
private void AddWord(TWord oNewWord)
{
// Use the Deskew method to use the same Y-Axis
TOCRRect oNewWordRect = Deskew(oNewWord.Rect);
// Build candidate lines
List <TLine> oCandidateLines = new List <TLine>();
foreach (TLine oLine in m_oLines)
{
if (Deskew(oLine.Rect).Bottom + MMtoPixel(20) < oNewWordRect.Top)
{
continue;
}
if (Deskew(oLine.Rect).Top - MMtoPixel(20) > oNewWordRect.Bottom)
{
break;
}
oCandidateLines.Add(oLine);
}
// Try 1: Find closer matching on top
foreach (TLine oLine in oCandidateLines)
{
foreach (TWord oWord in oLine)
{
TOCRRect oWordRect = Deskew(oWord.Rect);
// If the word in
if (oWordRect.IsEqualTop(oNewWordRect))
{
oLine.AddWord(oNewWord);
return;
}
}
}
// Now we remove lines that one word is above or beneath the new word
List <TLine> oCandidateLines2 = new List <TLine>();
foreach (TLine oLine in oCandidateLines)
{
bool bLineOk = true;
foreach (TWord oWord in oLine)
{
TOCRRect oWordRect = Deskew(oWord.Rect);
// Word is above or beneath the new word
if ((oNewWordRect.Top > oWordRect.Bottom) || (oNewWordRect.Bottom < oWordRect.Top))
{
bLineOk = false;
break;
}
}
if (bLineOk)
{
oCandidateLines2.Add(oLine);
}
}
// Try 2: Find closer matching on bottom
foreach (TLine oLine in oCandidateLines2)
{
foreach (TWord oWord in oLine)
{
TOCRRect oWordRect = Deskew(oWord.Rect);
// If the word in
if (oWordRect.IsEqualBottom(oNewWordRect))
{
oLine.AddWord(oNewWord);
return;
}
}
}
// Try 3: Find far matching from top
foreach (TLine oLine in oCandidateLines2)
{
foreach (TWord oWord in oLine)
{
TOCRRect oWordRect = Deskew(oWord.Rect);
if (Math.Abs(oNewWordRect.Top - oWordRect.Top) <=
Math.Min(MMtoPixel(2), oWordRect.Height / 2))
{
oLine.AddWord(oNewWord);
return;
}
}
}
// Try 4: Find far matching from bottom
foreach (TLine oLine in oCandidateLines2)
{
foreach (TWord oWord in oLine)
{
TOCRRect oWordRect = Deskew(oWord.Rect);
if (Math.Abs(oNewWordRect.Bottom - oWordRect.Bottom) <=
Math.Min(MMtoPixel(2), oWordRect.Height / 2))
{
oLine.AddWord(oNewWord);
return;
}
}
}
// We do not found any line that can take the word, so we create a new line
TLine oNewLine = new TLine();
oNewLine.AddWord(oNewWord);
AddLine(oNewLine);
}