lsm/PdfPig
1
0
Fork 0
mirror of https://github.com/UglyToad/PdfPig.git synced 2025-09-18 18:27:55 +08:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

fix Docstrum's GetTranslatedPoint() to handle dXj=0

Browse Source
This commit is contained in:
BobLd
2020-06-20 14:19:23 +01:00
parent 091c17bdf8
commit 924c0138e0
1 changed files with 14 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
src/UglyToad.PdfPig.DocumentLayoutAnalysis/PageSegmenter/DocstrumBoundingBoxes.cs
View File

@@ -9,7 +9,6 @@
using System.Threading.Tasks; using System.Threading.Tasks;
using UglyToad.PdfPig.DocumentLayoutAnalysis.ReadingOrderDetector; using UglyToad.PdfPig.DocumentLayoutAnalysis.ReadingOrderDetector;
/// <inheritdoc />
/// <summary> /// <summary>
/// The Document Spectrum (Docstrum) algorithm is a bottom-up page segmentation technique based on nearest-neighbourhood /// The Document Spectrum (Docstrum) algorithm is a bottom-up page segmentation technique based on nearest-neighbourhood
/// clustering of connected components extracted from the document. /// clustering of connected components extracted from the document.
@@ -23,7 +22,6 @@
/// </summary> /// </summary>
public static DocstrumBoundingBoxes Instance { get; } = new DocstrumBoundingBoxes(); public static DocstrumBoundingBoxes Instance { get; } = new DocstrumBoundingBoxes();
/// <inheritdoc />
/// <summary> /// <summary>
/// Get the blocks using default options values. /// Get the blocks using default options values.
/// </summary> /// </summary>
@@ -34,7 +32,6 @@
return GetBlocks(words, new DocstrumBoundingBoxesOptions()); return GetBlocks(words, new DocstrumBoundingBoxesOptions());
} }
/// <inheritdoc />
/// <summary> /// <summary>
/// Get the blocks using options values. /// Get the blocks using options values.
/// </summary> /// </summary>
@@ -320,7 +317,6 @@
/// <summary> /// <summary>
/// Perpendicular overlapping distance. /// Perpendicular overlapping distance.
/// TODO: describe checks done
/// </summary> /// </summary>
/// <param name="line1"></param> /// <param name="line1"></param>
/// <param name="line2"></param> /// <param name="line2"></param>
@@ -419,7 +415,6 @@
overlap = false; overlap = false;
} }
//double pj = Math.Sqrt((Dj.Y - Cj.Y) * (Dj.Y - Cj.Y) + (Dj.X - Cj.X) * (Dj.X - Cj.X));
double pj = Distances.Euclidean(Cj, Dj); double pj = Distances.Euclidean(Cj, Dj);
normalisedOverlap = (overlap ? pj : -pj) / j.Length; normalisedOverlap = (overlap ? pj : -pj) / j.Length;
@@ -448,21 +443,29 @@
double dYidYj = dYi * dYj; double dYidYj = dYi * dYj;
double dXidXj = dXi * dXj; double dXidXj = dXi * dXj;
double denominator = dYidYj + dXidXj; double denominator = dYidYj + dXidXj;
if (denominator.AlmostEqualsToZero(epsilon)) if (denominator.AlmostEqualsToZero(epsilon))
{ {
// The denominator is 0 when translating points, meaning the lines are perpendicular. // The denominator is 0 when translating points, meaning the lines are perpendicular.
return null; return null;
} }
double xTj = (xPi * dXidXj + xPj * dYidYj + dXj * dYi * (yPi - yPj)) / denominator; double xAj;
double yTj = yPj; // TODO: need to check that double yAj;
if (dXj > epsilon) if (!dXj.AlmostEqualsToZero(epsilon)) // dXj != 0
{ {
yTj = dYj / dXj * (xTj - xPj) + yPj; xAj = (xPi * dXidXj + xPj * dYidYj + dXj * dYi * (yPi - yPj)) / denominator;
yAj = dYj / dXj * (xAj - xPj) + yPj;
}
else // If dXj = 0, then yAj is calculated first, and xAj is calculated from that.
{
// TODO: check that
yAj = (yPi * dYidYj + yPj * dXidXj + dYj * dXi * (xPi - xPj)) / denominator;
xAj = xPj;
} }
return new PdfPoint(xTj, yTj); return new PdfPoint(xAj, yAj);
} }
/// <summary> /// <summary>
@@ -482,8 +485,7 @@
double dotProd1 = ax * bx + ay * by; double dotProd1 = ax * bx + ay * by;
if (dotProd1 < 0) return false; if (dotProd1 < 0) return false;
double dotProd2 = bx * bx + by * by; return dotProd1 <= (bx * bx + by * by);
return dotProd1 <= dotProd2;
} }
/// <summary> /// <summary>