static void GetInsideLine(CellLines horizontalSamples, CellLines verticalSamples, int lineDistance, ref CellLines horizontals, ref CellLines verticals, out CellLines horizontalRests, out CellLines verticalRests)
{
horizontalRests = new CellLines();
verticalRests = new CellLines();
bool isContain;
// 가장 외각선과 맞닿는 부분이 있는 라인과 맞닿은 라인은 인정한다.
foreach (CellLine horizontal in horizontalSamples)
{
isContain = false;
foreach (CellLine vertical in verticals)
{
if (horizontal.StartY >= vertical.StartY && horizontal.EndY <= vertical.EndY)
{
if (Math.Abs(horizontal.StartX - vertical.StartX) <= lineDistance || Math.Abs(horizontal.EndX - vertical.EndX) <= lineDistance)
{
isContain = true;
horizontals.Add(horizontal);
break;
}
}
}
if (!isContain)
{
horizontalRests.Add(horizontal);
}
}
// 가장 외각선과 맞닿는 부분이 있는 라인과 맞닿은 라인은 인정한다.
foreach (CellLine vertical in verticalSamples)
{
isContain = false;
foreach (CellLine horizontal in horizontals)
{
if (vertical.StartX >= horizontal.StartX && vertical.EndX <= horizontal.EndX)
{
if (Math.Abs(vertical.StartY - horizontal.StartY) <= lineDistance || Math.Abs(vertical.EndY - horizontal.EndY) <= lineDistance)
{
isContain = true;
verticals.Add(vertical);
break;
}
}
}
if (!isContain)
{
verticalRests.Add(vertical);
}
}
}
static CellLines GetVerticalAllInBox(CellLine top, CellLine bottom, CellLine left, CellLine right, CellLines verticals, int lineDistance, ref Dictionary <int, int> duplicateDic)
{
CellLines verticalsAll = new CellLines();
foreach (CellLine line in verticals)
{
if (!duplicateDic.ContainsKey(line.Index) &&
line.StartX >= left.StartX &&
line.EndX <= right.EndX &&
line.StartY >= top.StartY - lineDistance &&
line.EndY <= bottom.StartY + lineDistance)
{
verticalsAll.Add(line);
}
}
return(verticalsAll);
}
static CellLines DetectHorizontalLines(Mat binary, int hThreshold, int textSize)
{
CellLines horizontals = new CellLines();
using (Mat horizontal = binary.Clone())
using (Mat hKernel = Cv2.GetStructuringElement(shape: MorphShapes.Rect, ksize: new Size(hThreshold, SIZE_LINE)))
using (Mat kernelDilate = Cv2.GetStructuringElement(shape: MorphShapes.Rect, ksize: new Size(SIZE_DILATE, SIZE_DILATE)))
{
// horizontal
Cv2.Erode(src: horizontal, dst: horizontal, element: hKernel, anchor: new Point(-1, -1));
Cv2.Dilate(src: horizontal, dst: horizontal, element: hKernel, anchor: new Point(-1, -1));
// 중간에 살짝 끊어진 라인을 잇기 위해 라인을 확장시킨다.
Cv2.Dilate(src: horizontal, dst: horizontal, element: kernelDilate, anchor: new Point(-1, -1));
Point[][] horizontalContours;
HierarchyIndex[] horizontalHierarchy;
Cv2.FindContours(image: horizontal, contours: out horizontalContours, hierarchy: out horizontalHierarchy, mode: RetrievalModes.External, method: ContourApproximationModes.ApproxSimple, offset: new Point(0, 0));
int startX, startY, endX, endY, index = 0;
Rect rect;
for (int i = 0; i < horizontalHierarchy.Length; i++)
{
rect = Cv2.BoundingRect(curve: horizontalContours[i]);
startX = rect.X;
startY = rect.Y + (int)(rect.Height * 0.5);
endX = rect.X + rect.Width;
endY = startY;
if (rect.Width > textSize)
{
horizontals.Add(new CellLine(index: index++, startX: startX, startY: startY, endX: endX, endY: endY, thickness: rect.Height));
}
}
}
return(new CellLines(horizontals.OrderBy(line => line.CenterY)));
}
static void RemoveNoiseLine(CellLine top, CellLine bottom, CellLine left, CellLine right, CellLines innerHorizontals, CellLines innerVerticals, int lineDistance, ref Dictionary <int, int> duplicateDic, out CellLines horizontalsFinal, out CellLines verticalsFinal)
{
// 최종적으로 찾은 것을 duplicateDic에 추가한다. (이렇게 하면 박스 안의 박스는 별도로 처리 가능)
horizontalsFinal = new CellLines();
verticalsFinal = new CellLines();
CellLines horizontals = new CellLines();
CellLines verticals = new CellLines();
CellLines horizontalFirst = new CellLines();
CellLines verticalFirst = new CellLines();
// 일단 가장 외각선과 맞닿는 부분이 있는 라인만 인정한다.
foreach (CellLine line in innerHorizontals)
{
if (Math.Abs(line.StartX - left.StartX) <= lineDistance || Math.Abs(line.EndX - right.EndX) <= lineDistance)
{
horizontals.Add(line);
}
else
{
horizontalFirst.Add(line);
}
}
// 일단 가장 외각선과 맞닿는 부분이 있는 라인만 인정한다.
foreach (CellLine line in innerVerticals)
{
if (Math.Abs(line.StartY - top.StartY) <= lineDistance || Math.Abs(line.EndY - bottom.EndY) <= lineDistance)
{
verticals.Add(line);
}
else
{
verticalFirst.Add(line);
}
}
// 가장 외각선과 맞닿는 부분이 있는 라인과 맞닿은 라인은 인정한다.
// Data Sheet 같이 복잡한 테이블을 처리하기 위해 4번 돌린다.
CellLines horizontalSecond, verticalSecond, horizontalThird, verticalThird, horizontalFourth, verticalFourth;
GetInsideLine(horizontalSamples: horizontalFirst, verticalSamples: verticalFirst, lineDistance: lineDistance, horizontals: ref horizontals, verticals: ref verticals, horizontalRests: out horizontalSecond, verticalRests: out verticalSecond);
GetInsideLine(horizontalSamples: horizontalSecond, verticalSamples: verticalSecond, lineDistance: lineDistance, horizontals: ref horizontals, verticals: ref verticals, horizontalRests: out horizontalThird, verticalRests: out verticalThird);
GetInsideLine(horizontalSamples: horizontalThird, verticalSamples: verticalThird, lineDistance: lineDistance, horizontals: ref horizontals, verticals: ref verticals, horizontalRests: out horizontalFourth, verticalRests: out verticalFourth);
bool matchStart, matchEnd;
// top, bottom을 수평선에 추가한다.
horizontals.Add(top);
horizontals.Add(bottom);
// left, right를 수직선에 추가한다.
verticals.Add(left);
verticals.Add(right);
// 양 끝점이 모두 vertical과 붙어 있어야 line으로 인정한다. Data Sheet의 경우 단순 밑줄이 마치 Line처럼 잡힘
foreach (CellLine horizontal in horizontals)
{
matchStart = false;
matchEnd = false;
foreach (CellLine vertical in verticals)
{
if (horizontal.StartY >= vertical.StartY && horizontal.EndY <= vertical.EndY)
{
if (Math.Abs(horizontal.StartX - vertical.StartX) <= lineDistance)
{
matchStart = true;
}
else if (Math.Abs(horizontal.EndX - vertical.EndX) <= lineDistance)
{
matchEnd = true;
}
}
if (matchStart && matchEnd)
{
horizontalsFinal.Add(horizontal);
// top, bottom이 추가 되었기 때문에 체크
if (!duplicateDic.ContainsKey(horizontal.Index))
{
duplicateDic.Add(horizontal.Index, horizontal.Index);
}
break;
}
}
}
// 양 끝점이 모두 horizontal과 붙어 있어야 line으로 인정한다. Data Sheet의 경우 단순 밑줄이 마치 Line처럼 잡힘
foreach (CellLine vertical in verticals)
{
matchStart = false;
matchEnd = false;
foreach (CellLine horizontal in horizontals)
{
if (vertical.StartX >= horizontal.StartX && vertical.EndX <= horizontal.EndX)
{
if (Math.Abs(vertical.StartY - horizontal.StartY) <= lineDistance)
{
matchStart = true;
}
else if (Math.Abs(vertical.EndY - horizontal.EndY) <= lineDistance)
{
matchEnd = true;
}
}
if (matchStart && matchEnd)
{
verticalsFinal.Add(vertical);
// left, right가 추가 되었기 때문에 체크
if (!duplicateDic.ContainsKey(vertical.Index))
{
duplicateDic.Add(vertical.Index, vertical.Index);
}
break;
}
}
}
}
