static ImageTableCollection DetectTables(CellBoxes cellBoxes, int pageWidth, int pageHeight, int lineDistance, int textSize)
{
ImageTableCollection tables = new ImageTableCollection();
CrossPoints crossPoints;
ImageCellCollection cells;
int[,] crossPointIndexes;
int index = 0, minX, minY, maxX, maxY;
int spacingMin = (int)(textSize * RATIO_TEXT_HEIGHT);
double widthRatio = 1d / pageWidth;
double heightRatio = 1d / pageHeight;
foreach (CellBox box in cellBoxes)
{
// box 안의 line이 5개가 넘어야 table로 인정한다. top, bottom, left, right가 있으므로 나머지 line이 적어도 1개 이상 있어야 함.
if (box.Horizontals.Count + box.Verticals.Count > 0)
{
// 교차점을 구한다.
crossPoints = DetectRawCrossPoints(horizontals: box.Horizontals, verticals: box.Verticals, lineDistance: lineDistance, crossPointIndexes: out crossPointIndexes);
// 일단 cell을 구한다.
cells = DetectRawCells(crossPoints: crossPoints, crossPointIndexes: crossPointIndexes, spacingMin: spacingMin, widthRatio: widthRatio, heightRatio: heightRatio);
if (cells.Count > 0)
{
minX = minY = int.MaxValue;
maxX = maxY = int.MinValue;
foreach (ImageCell cell in cells)
{
if (cell.LeftX < minX)
{
minX = cell.LeftX;
}
if (cell.RightX > maxX)
{
maxX = cell.RightX;
}
if (cell.TopY < minY)
{
minY = cell.TopY;
}
if (cell.BottomY > maxY)
{
maxY = cell.BottomY;
}
}
tables.Add(new ImageTable(index: index++, x: minX, y: minY, width: maxX - minX, height: maxY - minY, cells: cells));
}
}
}
return(tables);
}
public static ImageTableCollection GetTables(string imagePath, int textSize, bool drawTable = false, bool drawLine = false)
{
CellLines horizontals, verticals;
int pageWidth, pageHeight;
DetectLines(imagePath: imagePath, textSize: textSize, horizontals: out horizontals, verticals: out verticals, pageWidth: out pageWidth, pageHeight: out pageHeight);
int lineDistance = GetLineDistanceByModeThickness(horizontals: ref horizontals, verticals: ref verticals);
CellBoxes cellBoxes = DetectCellBoxes(lineDistance: lineDistance, horizontals: ref horizontals, verticals: ref verticals);
ImageTableCollection tables = DetectTables(cellBoxes: cellBoxes, pageWidth: pageWidth, pageHeight: pageHeight, lineDistance: lineDistance, textSize: textSize);
if (drawLine)
{
DrawLines(horizontals: horizontals, verticals: verticals, imagePath: imagePath);
}
if (drawTable)
{
DrawTables(tables: tables, imagePath: imagePath);
}
return(tables);
}
static CellBoxes DetectCellBoxes(int lineDistance, ref CellLines horizontals, ref CellLines verticals)
{
CellBoxes cellBoxes = new CellBoxes();
CellLines horizontalsFinal, verticalsFinal;
CellLine left = new CellLine();
CellLine right = new CellLine();
CellLine top, bottom;
bool findLeft, findRight;
int index = 0;
Dictionary <int, int> duplicateDic = new Dictionary <int, int>();
// 위에서부터 아래로 찾는다.
foreach (CellLine horizontal in horizontals)
{
if (!duplicateDic.ContainsKey(horizontal.Index))
{
findLeft = findRight = false;
foreach (CellLine vertical in verticals)
{
if (!duplicateDic.ContainsKey(vertical.Index))
{
if (horizontal.MinX <= vertical.MaxX && horizontal.MaxX >= vertical.MinX && horizontal.MinY <= vertical.MaxY && horizontal.MaxY >= vertical.MinY)
{
// 현재 라인의 왼쪽과 같은 위치에서 시작되는 수직 라인을 찾는다.
if (Math.Abs(horizontal.StartX - vertical.StartX) <= lineDistance)
{
findLeft = true;
left = vertical;
}
// 현재 라인의 오른쪽과 같은 위치에서 시작되는 수직 라인을 찾는다.
else if (Math.Abs(horizontal.EndX - vertical.StartX) <= lineDistance)
{
findRight = true;
right = vertical;
}
}
// 두 라인을 찾으면
if (findLeft && findRight)
{
// 두 라인을 이용해서 두 수직 라인의 하단과 이어지는 수평 라인을 찾는다.
if (ExistBottomLine(topIndex: horizontal.Index, lineDistance: lineDistance, verticalLeft: left, verticalRight: right, horizontals: horizontals, duplicateDic: ref duplicateDic, bottom: out bottom))
{
top = horizontal;
duplicateDic.Add(top.Index, top.Index);
duplicateDic.Add(bottom.Index, bottom.Index);
duplicateDic.Add(left.Index, left.Index);
duplicateDic.Add(right.Index, right.Index);
horizontals = GetHorizonsAllInBox(lineDistance: lineDistance, top: top, bottom: bottom, left: left, right: right, horizontals: horizontals, duplicateDic: ref duplicateDic);
verticals = GetVerticalAllInBox(lineDistance: lineDistance, top: top, bottom: bottom, left: left, right: right, verticals: verticals, duplicateDic: ref duplicateDic);
RemoveNoiseLine(lineDistance: lineDistance, top: top, bottom: bottom, left: left, right: right, innerHorizontals: horizontals, innerVerticals: verticals, duplicateDic: ref duplicateDic, horizontalsFinal: out horizontalsFinal, verticalsFinal: out verticalsFinal);
cellBoxes.Add(new CellBox(index: index++, horizontals: new CellLines(horizontalsFinal.OrderBy(line => line.StartY)), verticals: new CellLines(verticalsFinal.OrderBy(line => line.StartX))));
break;
}
}
}
}
}
}
return(cellBoxes);
}
