using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using OpenCvSharp;
namespace TargetMeasureObject
{
class GeometricProcessor:NormalProcessor
{
static public double ContourBaryCenterOrientationGet(Point[] cnt)
{
var minAreaRect = Cv2.MinAreaRect(cnt);
Point2f[] minAreaRectPts = Cv2.BoxPoints(minAreaRect);
var M = Cv2.Moments(cnt);
Point baryCenter = new Point((int)(M.M10 / M.M00), (int)(M.M01 / M.M00));
Func<Point2f, Point2f, double> PpDistance = (p1, p2) =>
{
return Math.Sqrt(Math.Pow((p1.X - p2.X), 2) + Math.Pow((p1.Y - p2.Y), 2));
};
Func<Point2f, Point2f, Point2f,Point2f, double> PpWidAngle = (p0, p1, p2,p3) =>
{
double angle = double.NaN;
Point2f pd = new Point2f();
Point2f pu = new Point2f();
Point2f pRemain = new Point2f();
if (PpDistance(p0, p1) > PpDistance(p1, p2))
{
pd = p0;
}
pd = (PpDistance(p0, p1) > PpDistance(p1, p2)) ? p1 : p1;
pu = (PpDistance(p0, p1) > PpDistance(p1, p2)) ? p0 : p2;
pRemain = pu == p0 ? p2 : p0;
double areaDown = Cv2.ContourArea(new Point[] { (Point)pd, (Point)pRemain, baryCenter });
double areaUp = Cv2.ContourArea(new Point[] { (Point)pu, (Point)p3, baryCenter });
angle = Math.Atan((pu.Y - pd.Y) / (pu.X - pd.X)) * (180 / Math.PI);
if ((int)PpDistance(p0, p1) != (int)PpDistance(p1, p2))
{
if (angle == 0 || angle == 90)
{
angle = angle == 0 ? areaDown >= areaUp ? 0 : 180 : areaDown >= areaUp ? 90 : 270;
}
else
{
if (areaDown >= areaUp)
{
if (angle > 0 && angle < 90)
{
angle = 180 - angle;
}
else if (angle < 0)
{
angle = Math.Abs(angle);
}
}
else
{
if (angle > 0 && angle < 90)
{
angle = 360 - angle;
}
else if (angle < 0 && angle > -90)
{
angle = Math.Abs(angle) + 180;
}
}
}
}
else
{
angle = angle == 90 ? 0 : angle;
}
return angle;
};
return PpWidAngle(minAreaRectPts[0], minAreaRectPts[1], minAreaRectPts[2], minAreaRectPts[3]);
}
/// <summary>
/// 获取轮廓最大宽度得方向
/// </summary>
/// <param name="minAreaRectPts">轮廓最小面积矩形boxpoints</param>
/// <returns>最大宽度得角度</returns>
static public void ContourWidthOrientionGet(Point2f[] minAreaRectPts)
{
//Func<Point2f, Point2f, double> PpDistance = (p1, p2) =>
//{
// return Math.Sqrt(Math.Pow((p1.X - p2.X), 2) + Math.Pow((p1.Y - p2.Y), 2));
//};
//Func<Point2f, Point2f, Point2f, double> PpWidAngle = (p0, p1, p2) =>
//{
// double angle = double.NaN;
// Point2f pd = new Point2f();
// Point2f pu = new Point2f();
// if (PpDistance(p0, p1) > PpDistance(p1, p2))
// {
// pd = p0;
// }
// pd = (PpDistance(p0, p1) > PpDistance(p1, p2)) ? p1 : p1;
// pu = (PpDistance(p0, p1) > PpDistance(p1, p2)) ? p0 : p2;
// angle = Math.Atan((pu.Y - pd.Y) / (pu.X - pd.X)) * (180 / Math.PI);
// return angle;
//};
//return PpWidAngle(minAreaRectPts[0], minAreaRectPts[1], minAreaRectPts[2]);
}
/// <summary>
/// 通过几何特征将待处理图像处理成模板宽度方向得图像
/// </summary>
/// <param name="pattern"></param>
/// <param name="other"></param>
/// <param name="deltImg"></param>
/// <returns></returns>
static public Mat ContourWidthOrientionRotation(ImageGeometricData pattern, ImageGeometricData other,Mat deltImg,bool show=false)
{
Mat _outImg = deltImg.Clone();
double deltaAng = pattern.ContourWidthOriention - other.ContourWidthOriention;
_outImg = ImageRotate(_outImg, new Point2f(_outImg.Width / 2, _outImg.Height / 2), -deltaAng);
if (show)
{
ImageShow("ContourWidthOrientionDealShowing", ImagesMerge(new Mat[] { deltImg }, _outImg));
}
return _outImg;
}
/// <summary>
/// 通过提供几何数据中该几何边界矩形为参考，对待处理图像得边界矩形进行缩放
/// </summary>
/// <param name="pattern">模板几何参数</param>
/// <param name="deltImg">待处理图像（二值图）</param>
/// <param name="show"></param>
/// <returns></returns>
static public Mat ContourSizeStretch(ImageGeometricData pattern,Mat deltImg, bool show=false)
{
Mat _tImg = deltImg.Clone();
Rect deltRect= BlobProcessor.ImageConnectedFieldSegment(_tImg,show:false).ConnectedFieldDatas.FieldRects[0];
double maxDiffRatio =
Math.Abs(pattern.ContourBoundRect.Width - deltRect.Width) >
Math.Abs(pattern.ContourBoundRect.Height - deltRect.Height) ?
(double)pattern.ContourBoundRect.Width / (double)deltRect.Width :
(double)pattern.ContourBoundRect.Height / (double)deltRect.Height;
Cv2.Resize(_tImg, _tImg, new Size(0, 0), maxDiffRatio, maxDiffRatio, InterpolationFlags.Cubic);
if (show)
{
ImageShow("ContourSizeDealShowing", ImagesMerge(new Mat[] { deltImg }, _tImg));
}
return _tImg;
}
/// <summary>
/// 获取两个轮廓尺寸相同的几何特征匹配成绩，其中轮廓数量权重较高，外部轮廓点数量轮廓较低，其他平均
/// </summary>
/// <param name="geo1"></param>
/// <param name="geo2"></param>
/// <param name="matchType"></param>
/// <returns></returns>
static public ImageGeometricScore GeometricsScoreCalc(ImageGeometricData geo1, ImageGeometricData geo2, ENUMS.GEOMETRIC_MATCH_TYPE matchType)
{
ImageGeometricScore gScore = new ImageGeometricScore();
double score = 0;
double scoreSum = 0;
bool frog = false;
bool flag = matchType is ENUMS.GEOMETRIC_MATCH_TYPE.ALL;
switch (matchType)
{
case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_AREA:
{
score = (geo1.ContourArea > geo2.ContourArea) ? geo2.ContourArea / geo1.ContourArea : geo1.ContourArea / geo2.ContourArea;
gScore.ContourAreaScore=score *= 100;
if (frog)
{
scoreSum += score;
goto case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_LENGTH;
}
break;
}
case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_LENGTH:
{
score = (geo1.ContourLength > geo2.ContourLength) ? geo2.ContourLength / geo1.ContourLength : geo1.ContourLength / geo2.ContourLength;
gScore.ContourLengthScore = score *= 100;
if (frog)
{
scoreSum += score;
goto case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_AREA_LEN_RATIO;
}
break;
}
case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_AREA_LEN_RATIO:
{
score = (geo1.ContourALRatio > geo2.ContourALRatio) ? geo2.ContourALRatio / geo1.ContourALRatio : geo1.ContourALRatio / geo2.ContourALRatio;
gScore.ContourALRatioScore = score *= 100;
if (frog)
{
scoreSum += score;
goto case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_NUMBER;
}
break;
}
case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_NUMBER:
{
//当轮廓数量不相等时，成绩为0
score = (geo1.ContourNumber != geo2.ContourNumber) ? 0: 1;
gScore.ContourNumberScore = score *= 130;
if (frog)
{
scoreSum += score;
goto case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_OUT_POINT;
}
break;
}
case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_OUT_POINT:
{
/*经过测试发现，当图像处理效果较差时，外部轮廓点数量对检测效果并不明显，因此缩小
外部轮廓点数量得权重*/
score = (geo1.ContourOutPtsNumber > geo2.ContourOutPtsNumber) ? ((double)geo2.ContourOutPtsNumber / (double)geo1.ContourOutPtsNumber) : ((double)geo1.ContourOutPtsNumber / (double)geo2.ContourOutPtsNumber);
gScore.ContourOutPtsNumberScore = score *= 70;
if (frog)
{
scoreSum += score;
goto case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_MINAREARL_LEN_RATIO;
}
break;
}
case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_MINAREARL_LEN_RATIO:
{
score = (geo1.ContourLMinRectRatio > geo2.ContourLMinRectRatio) ? geo2.ContourLMinRectRatio / geo1.ContourLMinRectRatio : geo1.ContourLMinRectRatio / geo2.ContourLMinRectRatio;
gScore.ContourLMinRectRatioScore = score *= 100;
if (frog)
{
scoreSum += score;
goto case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_MINAREARC_AREA_RATIO;
}
break;
}
case ENUMS.GEOMETRIC_MATCH_TYPE.CONTOUR_MINAREARC_AREA_RATIO:
{
score = (geo1.ContourAMinCircleRatio > geo2.ContourAMinCircleRatio) ? geo2.ContourAMinCircleRatio / geo1.ContourAMinCircleRatio : geo1.ContourAMinCircleRatio / geo2.ContourAMinCircleRatio;
gScore.ContourAMinCircleRatioScore = score *= 100;
if (frog)
{
scoreSum += score;
flag = false;
goto case ENUMS.GEOMETRIC_MATCH_TYPE.ALL;
}
break;
}
case ENUMS.GEOMETRIC_MATCH_TYPE.ALL:
{
if (flag)
{
frog = true;
goto case (ENUMS.GEOMETRIC_MATCH_TYPE)0;
}
scoreSum /= (double)(ENUMS.GEOMETRIC_MATCH_TYPE.ALL);
gScore.AnverageScore = scoreSum;
break;
}
}
return gScore;
}
static public ImageGeometricData GeometricInformGet(Mat img, double thresh=120, bool show = true)
{
ImageGeometricData ImageGeometricData = new ImageGeometricData();
using (var uImg = img.Clone())
{
Cv2.Threshold(uImg, uImg, thresh, 255, ThresholdTypes.Binary);
Cv2.FindContours(uImg, out Point[][] cnts, out HierarchyIndex[] hIndexs, RetrievalModes.External, ContourApproximationModes.ApproxSimple);
/*筛选轮廓中最大轮廓*/
cnts = cnts.ToList().OrderByDescending(cnt => Cv2.ContourArea(cnt)).ToArray();
double area = Cv2.ContourArea(cnts[0]);
double length = Cv2.ArcLength(cnts[0], true);
double ratio = area / length;
double epsilon = 2;
Cv2.MinEnclosingCircle(cnts[0], out Point2f center, out float radius);
double encloseCircleArea = Math.PI * (Math.Pow(radius, 2));
Rect boundRect = Cv2.BoundingRect(cnts[0]);
RotatedRect minARect = Cv2.MinAreaRect(cnts[0]);
/*获取轮廓宽度方向*/
double widOtiention = ContourBaryCenterOrientationGet(cnts[0]);
/*获取轮廓宽度方向*/
double minARectLen = (minARect.Size.Width + minARect.Size.Height) * 2;
Point[] approxCnts = Cv2.ApproxPolyDP(cnts[0], epsilon, true);
int ptsnum = approxCnts.Length;
/*获取全部轮廓中有效轮廓=>面积！=0 && 点数>2*/
Cv2.FindContours(uImg, out Point[][] aCnts, out HierarchyIndex[] nIndexs, RetrievalModes.List, ContourApproximationModes.ApproxSimple);
List<Point[]> nCnts = new List<Point[]>();
for(int i = 0; i < aCnts.Length; i++)
{
double _area = Math.Abs(Cv2.ContourArea(aCnts[i]));
if(aCnts[i].Length>2 && _area>5)
{
nCnts.Add(aCnts[i]);
}
}
int cntsNum = nCnts.Count;
ImageGeometricData.ContourArea = area;
ImageGeometricData.ContourLength = length;
ImageGeometricData.ContourALRatio = ratio;
double minAreaCirArea = Math.PI * Math.Pow(radius, 2);
//当轮廓拟合圆面积与轮廓面试比值超过0.9时，则该轮廓为圆
ImageGeometricData.ContourOutPtsNumber = area/ minAreaCirArea > 0.90?int.MaxValue:ptsnum;
ImageGeometricData.ContourNumber = cntsNum;
ImageGeometricData.ContourWidthOriention = widOtiention;
ImageGeometricData.ContourTotalPoints = nCnts.ToArray();
ImageGeometricData.ContourExternalPoints = cnts;
ImageGeometricData.ContourBoundRect = boundRect;
ImageGeometricData.ContourMinAreaRectWHRation = minARect.Size.Width > minARect.Size.Height ? (double)minARect.Size.Height / (double)minARect.Size.Width : (double)minARect.Size.Width / (double)minARect.Size.Height;
ImageGeometricData.ContourMinAreaRect = minARect;
ImageGeometricData.ContourLMinRectRatio = (minARectLen > length) ? length / minARectLen : minARectLen / length;
ImageGeometricData.ContourAMinCircleRatio = (area > encloseCircleArea) ? encloseCircleArea / area : area / encloseCircleArea;
ImageGeometricData.ContourApproxPoints = approxCnts;
if (show)
{
Mat _showImg = uImg.Clone();
Cv2.Merge(new Mat[] { uImg, uImg, uImg }, _showImg);
Cv2.DrawContours(_showImg, cnts, -1, Scalar.Blue, thickness: 2);
ImageShow("GeometricInformGet", _showImg);
}
return ImageGeometricData;
}
}
static public ImageGeometricData GeometricInformGet(Point[][] contour, bool show = true)
{
ImageGeometricData ImageGeometricData = new ImageGeometricData();
contour = contour.ToList().OrderByDescending(cnt => Cv2.ContourArea(cnt)).ToArray();
double area = Cv2.ContourArea(contour[0]);
double length = Cv2.ArcLength(contour[0], true);
double ratio = area / length;
double epsilon = 2;
Cv2.MinEnclosingCircle(contour[0], out Point2f center, out float radius);
double encloseCircleArea = Math.PI * (Math.Pow(radius, 2));
Rect boundRect = Cv2.BoundingRect(contour[0]);
RotatedRect minARect = Cv2.MinAreaRect(contour[0]);
/*获取轮廓宽度方向*/
double widOtiention = ContourBaryCenterOrientationGet(contour[0]);
/*获取轮廓宽度方向*/
double minARectLen = (minARect.Size.Width + minARect.Size.Height) * 2;
Point[] approxCnts = Cv2.ApproxPolyDP(contour[0], epsilon, true);
int ptsnum = approxCnts.Length;
/*获取全部轮廓中有效轮廓=>面积！=0 && 点数>2*/
List<Point[]> nCnts = new List<Point[]>();
for (int i = 0; i < contour.Length; i++)
{
double _area = Math.Abs(Cv2.ContourArea(contour[i]));
if (contour[i].Length > 2 && _area > 5)
{
nCnts.Add(contour[i]);
}
}
int cntsNum = nCnts.Count;
ImageGeometricData.ContourArea = area;
ImageGeometricData.ContourLength = length;
ImageGeometricData.ContourALRatio = ratio;
double minAreaCirArea = Math.PI * Math.Pow(radius, 2);
//当轮廓拟合圆面积与轮廓面试比值超过0.9时，则该轮廓为圆
ImageGeometricData.ContourOutPtsNumber = area / minAreaCirArea > 0.90 ? int.MaxValue : ptsnum;
ImageGeometricData.ContourNumber = cntsNum;
ImageGeometricData.ContourWidthOriention = widOtiention;
ImageGeometricData.ContourTotalPoints = nCnts.ToArray();
ImageGeometricData.ContourExternalPoints = new Point[][] { contour [0]};
ImageGeometricData.ContourBoundRect = boundRect;
ImageGeometricData.ContourMinAreaRectWHRation = minARect.Size.Width > minARect.Size.Height ? (double)minARect.Size.Height / (double)minARect.Size.Width : (double)minARect.Size.Width / (double)minARect.Size.Height;
ImageGeometricData.ContourMinAreaRect = minARect;
ImageGeometricData.ContourLMinRectRatio = (minARectLen > length) ? length / minARectLen : minARectLen / length;
ImageGeometricData.ContourAMinCircleRatio = (area > encloseCircleArea) ? encloseCircleArea / area : area / encloseCircleArea;
ImageGeometricData.ContourApproxPoints = approxCnts;
if (show)
{
Mat _showImg = new Mat(boundRect.Size,MatType.CV_8UC3,Scalar.Black);
Cv2.DrawContours(_showImg, contour, -1, Scalar.Blue, thickness: 2);
ImageShow("GeometricInformGet", _showImg);
}
return ImageGeometricData;
}
static public ImageGeometricScore ImageGeometricMatcher(Mat patternImg,Mat trainImg,double thresh=110, ENUMS.GEOMETRIC_MATCH_TYPE matchType=ENUMS.GEOMETRIC_MATCH_TYPE.ALL,bool show=false)
{
ImageGeometricScore result = new ImageGeometricScore();
if (patternImg.Channels() != trainImg.Channels())
{
ImageWarningShow("GeometricMatcher:Channel xx");
return null;
}
Mat _pImg = patternImg.Clone();
Mat _tImg = trainImg.Clone();
Cv2.Threshold(_pImg, _pImg, thresh, 255, ThresholdTypes.Binary);
Cv2.Threshold(_tImg, _tImg, thresh, 255, ThresholdTypes.Binary);
ImageGeometricData geo1 = GeometricInformGet(_pImg, show: false);
ImageGeometricData geo2 = GeometricInformGet(_tImg, show: false);
if (geo1.ContourMinAreaRectWHRation < 0.8 &&
geo2.ContourMinAreaRectWHRation < 0.8
&& Math.Abs(geo1.ContourWidthOriention - geo2.ContourWidthOriention) > 5)
{
/*模板宽高比大：模板角度与匹配图像角度不一致时，几何检测结果影响明显*/
_tImg = ContourWidthOrientionRotation(geo1, geo2, _tImg, show: false);
ImageBlobData tempbbData = BlobProcessor.ImageConnectedFieldSegment(_tImg, show: false);
_tImg = (Mat)tempbbData.ConnectedFieldDatas.SegmentedSubRegionImage[0].M2;
_tImg = ContourSizeStretch(geo1, _tImg, show: false);
ImagesMerge(new Mat[] { _pImg }, _tImg, show: false);
}
else
{
/*模板宽高比大较小且宽度角度差较小时，直接对二值图进行拉伸（按照较小图像尺寸）*/
ImageCoupleStretch(ref _pImg, ref _tImg, _pImg.Size() != _tImg.Size(), false, _show: false);
}
/*图像尺寸变化后出现细微差距，简易进行模糊处理*/
geo1 = GeometricInformGet(_pImg, show: false);
geo2 = GeometricInformGet(_tImg, show: false);
/*模板角度与匹配图像角度不一致时，测试结果影响明显*/
/*轮廓形状特征计算-计算结果不佳，匹配效果较差，此处不使用*/
{
//var score1 = Cv2.MatchShapes(geo1.ContourExternalPoints[0], geo2.ContourExternalPoints[0], ShapeMatchModes.I1);
//var score2 = Cv2.MatchShapes(geo1.ContourExternalPoints[0], geo2.ContourExternalPoints[0], ShapeMatchModes.I2);
//var score3 = Cv2.MatchShapes(geo1.ContourExternalPoints[0], geo2.ContourExternalPoints[0], ShapeMatchModes.I3);
//Mat _pCntImg = new Mat();
//Mat _tCntImg = new Mat();
//for (int i = 0; i < geo1.ContourExternalPoints[0].Length; i++)
//{
// _pCntImg.PushBack(geo1.ContourExternalPoints[0][i]);
//}
//for (int i = 0; i < geo2.ContourExternalPoints[0].Length; i++)
//{
// _tCntImg.PushBack(geo2.ContourExternalPoints[0][i]);
//}
//var extractor = ShapeContextDistanceExtractor.Create();
//float escore1 = extractor.ComputeDistance(_pCntImg, _tCntImg);
//var hauExtractor = HausdorffDistanceExtractor.Create();
//float hauScore = hauExtractor.ComputeDistance(_pCntImg, _tCntImg);
}
/*轮廓形状特征计算*/
result = GeometricsScoreCalc(geo1, geo2, matchType);
if (show)
{
Cv2.Merge(new Mat[] { _pImg, _pImg, _pImg }, _pImg);
Cv2.Merge(new Mat[] { _tImg, _tImg, _tImg }, _tImg);
Cv2.DrawContours(_pImg, geo1.ContourTotalPoints, -1, Scalar.Blue, thickness: 2);
Cv2.DrawContours(_tImg, geo2.ContourTotalPoints, -1, Scalar.Green, thickness: 2);
ImagesMerge(new Mat[] { _pImg, _tImg }, null, ENUMS.IMAGE_PERMUTATION_TYPE.HORIZONTAL, show:show);
}
return result;
}
static public object ImageGeometricClassified(List<Pair> pairs, double qualifiedScore = 90, bool show = false)
{
List<Pair> _imgs = new List<Pair>();
for (int i = 0; i < pairs.Count; i++)
{
_imgs.Add(new Pair(pairs[i].M1, ((Mat)pairs[i].M2).Clone()));
}
List<List<Pair>> _outImges = new List<List<Pair>>();
while (_imgs.Count != 0)
{
List<Pair> _group = new List<Pair>();
_group.Add(_imgs[0]);
_imgs.RemoveAt(0);
for (int i = 0; i < _imgs.Count; i++)
{
double score = ImageGeometricMatcher((Mat)_group[0].M2, (Mat)_imgs[i].M2, show: false).AnverageScore;
if (score > qualifiedScore)
{
_group.Add(_imgs[i]);
_imgs.RemoveAt(i);
i -= 1;
}
}
_group = _group.OrderBy(pic => ((Mat)pic.M2).Width * ((Mat)pic.M2).Height).ToList();
_outImges.Add(_group);
}
if (show)
{
Mat _showImg = new Mat();
for (int i = 0; i < _outImges.Count; i++)
{
Mat temp = ImagesMerge(PairsMatQueues(_outImges[i]), null, ENUMS.IMAGE_PERMUTATION_TYPE.HORIZONTAL, ENUMS.PLACE_ORIENTATION.LEFT);
_showImg = ImagesMerge(new Mat[] { temp }, _showImg, placeOiren: ENUMS.PLACE_ORIENTATION.DOWN);
}
ImageShow("ImagesHistgramClassified", _showImg);
}
return _outImges;
}
static public object ImageGeometricMatchTool(Mat patImg,Mat detImg,double qualifiedScore=90,bool show=false)
{
List<ImageGeometricMatchResult> result = new List<ImageGeometricMatchResult>();
Mat detectImg = detImg.Clone(), patternImg = patImg.Clone();
Mat _sImg = detectImg.Clone();
if (_sImg.Type() == MatType.CV_8UC1)
{
Cv2.Merge(new Mat[] { _sImg, _sImg, _sImg }, _sImg);
}
ImageBlobData patBbData = new ImageBlobData();
ImageBlobData detBbData = new ImageBlobData();
/***获取图像直方图数据-为去除图像背景做准备***/
patBbData = BlobProcessor.ImageHisgramDataGet(patternImg);
/***去除图像背景-通过直方图特征***/
BlobProcessor.ImageBackgroundClear(patternImg, ref patBbData, blurSize: 5, thresh: 100);
detBbData = BlobProcessor.ImageHisgramDataGet(detectImg);
BlobProcessor.ImageBackgroundClear(detectImg, ref detBbData, blurSize: 5, thresh: 100);
/***获取图像连通域信息-为图像分割准备***/
BlobProcessor.ImageConnectedFieldSegment(detectImg, ref detBbData, detBbData.HistgramPeakPixel[0] * 2.3, show: true);
BlobProcessor.ImageConnectedFieldSegment(patternImg, ref patBbData, patBbData.HistgramPeakPixel[0] * 2.3, show: true);
Dictionary<int, Scalar> colorPool = MatchColorMaker(patBbData.ConnectedFieldDatas.SegmentedSubRegionImage.Count);
/***模板内的子图分类***/
var patGroups = (List<List<Pair>>)GeometricProcessor.ImageGeometricClassified(patBbData.ConnectedFieldDatas.SegmentedSubRegionImage, show: false) ;
for (int i = 0; i < detBbData.ConnectedFieldDatas.SegmentedSubRegionImage.Count; i++)
{
Mat _dImg = ((Mat)detBbData.ConnectedFieldDatas.SegmentedSubRegionImage[i].M2).Clone();
Rect _dRect = detBbData.ConnectedFieldDatas.FieldRects[i];
Rect _pRect = new Rect();
ImageGeometricScore bestScore = new ImageGeometricScore();
int bestID = -1;
for (int j = 0; j < patGroups.Count; j++)
{
Mat _pImg = ((Mat)patGroups[j][0].M2).Clone();
if(j==1 && i == 2)
{
}
var score = ImageGeometricMatcher(_pImg, _dImg, show: true);
if (score.AnverageScore > bestScore.AnverageScore)
{
bestScore = score;
_pRect = patBbData.ConnectedFieldDatas.FieldRects[(int)patGroups[j][0].M1];
bestID = j;
}
}
if (bestScore.AnverageScore > qualifiedScore)
{
ImageGeometricMatchResult _res = new ImageGeometricMatchResult();
_res.PatternID = bestID;
_res.MatchImages = new Pair(
((Mat)patBbData.ConnectedFieldDatas.SegmentedSubRegionImage[(int)patGroups[bestID][0].M1].M2).Clone(),
((Mat)detBbData.ConnectedFieldDatas.SegmentedSubRegionImage[i].M2).Clone());
_res.MatchPosition = new Pair(_pRect, _dRect);
_res.MatchScore = bestScore;
result.Add(_res);
_sImg.Rectangle(_dRect, colorPool[bestID], 2);
_sImg = ImageAnnotation(_sImg,
new List<List<string>>() { new List<string>() { "ModelID:"+bestID.ToString(),
"GeoScore:"+bestScore.AnverageScore.ToString("0.00"), } },
new List<Point>() { new Point(_dRect.X, _dRect.Y) });
}
}
if (show)
{
Mat _patAttach = new Mat();
for (int i = 0; i < patGroups.Count; i++)
{
_patAttach = ImagesMerge(new Mat[] { (Mat)patGroups[i][0].M2 }, _patAttach, placeOiren: ENUMS.PLACE_ORIENTATION.DOWN);
}
_sImg = ImagesMerge(new Mat[] { _patAttach }, _sImg);
ImageShow("sss", _sImg);
}
return result;
}
static public List<ImageEdgePtInform> ImageEdgeInformationGet(Mat img,double minContrast,double maxContrast,ref Point relaOrgPoint,ref Size edgeSize)
{
Mat _uImg = img.Clone();
if (_uImg.Type() == MatType.CV_8UC3)
{
Cv2.CvtColor(_uImg, _uImg, ColorConversionCodes.BGR2GRAY);
}
List<ImageEdgePtInform> validEdgeInforms = new List<ImageEdgePtInform>();
int Height = _uImg.Rows;
int Width = _uImg.Cols;
int edgePtCount = 0;
bool flag = true;
Size validSize = new Size();
Mat xMagnitude = _uImg.Sobel(MatType.CV_64FC1, 1, 0, 3);//double-幅度y
Mat yMagnitude = _uImg.Sobel(MatType.CV_64FC1, 0, 1, 3);//double-幅度x
Mat normMagnitude = Mat.Zeros(_uImg.Size(), MatType.CV_64FC1);//double
Mat orgMagnitude = Mat.Zeros(_uImg.Size(), MatType.CV_64FC1);//double-幅度
Mat orientation = Mat.Zeros(_uImg.Size(), MatType.CV_32FC1);//int-方向
double maxMagnitude = double.MinValue;
relaOrgPoint = new Point(int.MaxValue,int.MaxValue);
unsafe
{
//收集模板幅度和角度
for(int row = 0; row < Height; row++)
{
double* _xRowPtr = (double*)xMagnitude.Ptr(row);
double* _yRowPtr = (double*)yMagnitude.Ptr(row);
double* _magnitudeRowPtr = (double*)orgMagnitude.Ptr(row);
int* _orientRowPtr = (int*)orientation.Ptr(row);
for (int col = 0; col < Width; col++)
{
double dx = _xRowPtr[col];
double dy = _yRowPtr[col];
double _mag = Math.Sqrt(dx * dx + dy * dy);
double _orient = Math.Atan(dy / dx) * (180.0 / Math.PI);
if(_orient > -22.5&& _orient <= 22.5)
{
_orient = 0;
}
else if(_orient>22.5 && _orient <= 67.5)
{
_orient = 45;
}
else if(_orient>67.5 || _orient <= -67.5)
{
_orient = 90;
}
else if(_orient>-67.5 && _orient < -22.5)
{
_orient = -45;
}
else
{
_orient = 0;
}
_magnitudeRowPtr[col] = _mag;
_orientRowPtr[col] = (int)_orient;
maxMagnitude = Math.Max(maxMagnitude, _mag);
}
}
//模板幅度非极大值抑制
normMagnitude = orgMagnitude.Clone();
for (int row = 0; row < Height; row++)
{
int* _oriRowPtr = (int*)orientation.Ptr(row);
double* _magGrayRowPtr = (double*)normMagnitude.Ptr(row);
double* _magRowPtr = (double*)orgMagnitude.Ptr(row);
for (int col = 0; col < Width; col++)
{
int cenOrient = _oriRowPtr[col];
double cenMag = _magRowPtr[col];
double leftMag = 0;
double rightMag = 0;
switch (cenOrient)
{
case 0:
{
if(col-1>=0 && col + 1 < Width)
{
leftMag = ((double*)orgMagnitude.Ptr(row))[col - 1];
rightMag = ((double*)orgMagnitude.Ptr(row))[col + 1];
}
break;
}
case 45:
{
if(col-1>=0 && col+1 <Width && row-1>=0 && row + 1 < Height)
{
leftMag = ((double*)orgMagnitude.Ptr(row + 1))[col - 1];
rightMag = ((double*)orgMagnitude.Ptr(row - 1))[col + 1];
}
break;
}
case 90:
{
if (row - 1 >= 0 && row + 1 < Height)
{
leftMag = ((double*)orgMagnitude.Ptr(row - 1))[col];
rightMag = ((double*)orgMagnitude.Ptr(row + 1))[col];
}
break;
}
case -45:
{
if (col - 1 >= 0 && col + 1 < Width && row - 1 >= 0 && row + 1 < Height)
{
leftMag = ((double*)orgMagnitude.Ptr(row - 1))[col - 1];
rightMag = ((double*)orgMagnitude.Ptr(row + 1))[col + 1];
}
break;
}
default:
break;
}
if(cenMag<leftMag || cenMag < rightMag || row==0 || row==Height-1 || col==0 || col==Width-1 )
{
_magGrayRowPtr[col] = 0;
}
else
{
_magGrayRowPtr[col] = 255.0 * (cenMag / maxMagnitude);
}
}
}
//模板滞后阈值
for (int row = 0; row < Height; row++)
{
double* magRowPtr = (double*)orgMagnitude.Ptr(row);
for (int col = 0; col < Width; col++)
{
flag = true;
double grayMag = ((double*)normMagnitude.Ptr(row))[col];
if (grayMag < maxContrast)
{
if (grayMag < minContrast)
{
magRowPtr[col] = 0;
flag = false;
}
else
{
if (col - 1 >= 0 && col + 1 < Width && row - 1 >= 0 && row + 1 < Height)
{
if(((double*)normMagnitude.Ptr(row-1))[col-1]<maxContrast&&
((double*)normMagnitude.Ptr(row))[col-1] < maxContrast&&
((double*)normMagnitude.Ptr(row+1))[col-1] < maxContrast&&
((double*)normMagnitude.Ptr(row-1))[col] < maxContrast&&
((double*)normMagnitude.Ptr(row+1))[col] < maxContrast&&
((double*)normMagnitude.Ptr(row-1))[col+1] < maxContrast&&
((double*)normMagnitude.Ptr(row))[col+1] < maxContrast&&
((double*)normMagnitude.Ptr(row+1))[col+1] < maxContrast)
{
magRowPtr[col] = 0;
flag = false;
}
}
else
{
magRowPtr[col] = 0;
flag = false;
}
}
}
double dx = ((double*)xMagnitude.Ptr(row))[col];
double dy = ((double*)yMagnitude.Ptr(row))[col];
double mag = Math.Sqrt(dx * dx + dy * dy);
int* orient = (int*)orientation.Ptr(row);
if (flag && (dx!=0 || dy!=0))
{
validSize.Width = Math.Max(col, validSize.Width);
validSize.Height = Math.Max(row, validSize.Height);
relaOrgPoint.X = Math.Min(col, relaOrgPoint.X);
relaOrgPoint.Y = Math.Min(row, relaOrgPoint.Y);
ImageEdgePtInform edgeInform = new ImageEdgePtInform();
edgeInform.DerivativeX = dx;
edgeInform.DerivativeY = dy;
edgeInform.Magnitude = 1.0/ mag;
edgeInform.Orientation = orient[col];
edgeInform.RelativePos = new Point(col, row);
validEdgeInforms.Add(edgeInform);
edgePtCount++;
}
}
}
}
for (int i = 0; i < validEdgeInforms.Count; i++)
{
img.Circle(validEdgeInforms[i].RelativePos, 1, Scalar.Red);
}
ImageShow("ksdkskdks", img);
for (int i = 0; i < edgePtCount; i++)
{
int x = validEdgeInforms[i].RelativePos.X;
int y = validEdgeInforms[i].RelativePos.Y;
validEdgeInforms[i].RelativePos = new Point(x - relaOrgPoint.X, y - relaOrgPoint.Y);
}
edgeSize = new Size(validSize.Width-relaOrgPoint.X,validSize.Height-relaOrgPoint.Y);
return validEdgeInforms;
}
static public List<ImageEdgePtInform> ImageContourEdgeInfomGet(Mat img,int margin,ref Size validSize,bool show=false)
{
Mat _uImg = img.Clone();
if (_uImg.Type() != MatType.CV_8UC1)
{
Cv2.CvtColor(_uImg, _uImg, ColorConversionCodes.BGR2GRAY);
}
int edgePointCount = 0;
Point relaOrgPt = new Point(int.MaxValue,int.MaxValue);
List<ImageEdgePtInform> resultEdgeInforms = new List<ImageEdgePtInform>();
Mat _blur = _uImg.GaussianBlur(new Size(3, 3), 0);
Mat _threImg= _blur.Threshold(120, 255, ThresholdTypes.BinaryInv);
_threImg.FindContours(out Point[][] cnts, out HierarchyIndex[] hids, RetrievalModes.List, ContourApproximationModes.ApproxNone);
cnts = cnts.Where(cnt => Cv2.ContourArea(cnt) > 3 && Cv2.ArcLength(cnt, false) > 3 && cnt.Length > 3).ToArray();
cnts=cnts.ToList().OrderByDescending(cnt => Cv2.ContourArea(cnt)).ToArray();
foreach (var cnt in cnts) edgePointCount += cnt.Length;
Mat xDerivative = _uImg.Sobel(MatType.CV_64FC1, 1, 0, 3);
Mat yDerivative = _uImg.Sobel(MatType.CV_64FC1, 0, 1, 3);
//临时测试
var M = Cv2.Moments(cnts[0]);
Point bcenter = new Point((int)(M.M10 / M.M00), (int)(M.M01 / M.M00));
unsafe
{
foreach(var cnt in cnts)
{
foreach(var pt in cnt)
{
ImageEdgePtInform ptInform = new ImageEdgePtInform();
int row = pt.Y, col = pt.X;
relaOrgPt.X = Math.Min(relaOrgPt.X, col);
relaOrgPt.Y = Math.Min(relaOrgPt.Y, row);
validSize.Width = Math.Max(validSize.Width, col);
validSize.Height = Math.Max(validSize.Height, row);
double dx = ((double*)xDerivative.Ptr(row))[col];
double dy = ((double*)yDerivative.Ptr(row))[col];
double mag = Math.Sqrt(dx * dx + dy * dy);
double barycentOrient = ContourBaryCenterOrientationGet(cnts[0]);
ptInform.DerivativeX = dx;
ptInform.DerivativeY = dy;
ptInform.Magnitude = (double)(1.0 / mag);
ptInform.RelativePos = pt;
ptInform.BarycentOrient = barycentOrient;
resultEdgeInforms.Add(ptInform);
}
}
foreach (var inf in resultEdgeInforms)
{
inf.RelativePos = new Point(inf.RelativePos.X - relaOrgPt.X, inf.RelativePos.Y - relaOrgPt.Y);
}
validSize = new Size(validSize.Width - relaOrgPt.X, validSize.Height - relaOrgPt.Y);
if (show)
{
Mat _sImg = img.Clone();
_sImg.DrawContours(cnts, -1, Scalar.Green, 1);
_sImg.Circle(relaOrgPt, 2, Scalar.Red);
_sImg.Rectangle(new Rect(relaOrgPt, validSize), Scalar.Blue, 1);
_sImg = ImageBasicLineDrawing(_sImg, bcenter, orientation: ENUMS.IMAGE_PERMUTATION_TYPE.HORIZONTAL);
_sImg = ImageBasicLineDrawing(_sImg, bcenter, orientation: ENUMS.IMAGE_PERMUTATION_TYPE.VERTICAL);
_sImg.Circle(bcenter, 5, Scalar.Blue, -1);
ImageShow("asdsad", _sImg);
}
}
return resultEdgeInforms;
}
static public double ImageEdgeMatch(Mat img,List<ImageEdgePtInform> queryEdgeInforms,
double minScore,double greediness, Size validSize, out Point conformPoints)
{
Mat _uImg = img.Clone();
if (_uImg.Type() == MatType.CV_8UC3)
{
Cv2.CvtColor(_uImg, _uImg, ColorConversionCodes.BGR2GRAY);
}
Cv2.GaussianBlur(_uImg, _uImg, new Size(3, 3), 0);
int Width = _uImg.Width;
int Height = _uImg.Height;
int queryCount = queryEdgeInforms.Count;
double partialScore = 0;
double resultScore = 0;
conformPoints =new Point();
unsafe
{
Mat txMagnitude = _uImg.Sobel(MatType.CV_64FC1, 1, 0, 3);
Mat tyMagnitude = _uImg.Sobel(MatType.CV_64FC1, 0, 1, 3);
Mat torgMagnitude = Mat.Zeros(_uImg.Size(), MatType.CV_64FC1);
double normMinScore = minScore / (double)queryCount;
double normGreediness = ((1 - greediness * minScore) / (1 - greediness)) / queryCount;
for(int row = 0; row < Height; row++)
{
double* xMag = (double*)txMagnitude.Ptr(row);
double* yMag = (double*)tyMagnitude.Ptr(row);
double* oMag = (double*)torgMagnitude.Ptr(row);
for (int col = 0; col < Width; col++)
{
double dx = xMag[col], dy = yMag[col];
double _mag = Math.Sqrt(dx * dx + dy * dy);
oMag[col] = _mag;
}
}
for(int row = 0; row < Height; row++)
{
for (int col = 0; col < Width; col++)
{
double sum = 0;
double corSum = 0;
bool flag = false;
for (int cn = 0; cn < queryCount; cn++)
{
int xoff = queryEdgeInforms[cn].RelativePos.X;
int yoff = queryEdgeInforms[cn].RelativePos.Y;
int relaX = xoff + col;
int relaY = yoff + row;
if (relaY >= Height || relaX >= Width)
{
continue;
}
double txD = ((double*)txMagnitude.Ptr(relaY))[relaX];
double tyD = ((double*)tyMagnitude.Ptr(relaY))[relaX];
double tMag = ((double*)torgMagnitude.Ptr(relaY))[relaX];
double qxD = queryEdgeInforms[cn].DerivativeX;
double qyD = queryEdgeInforms[cn].DerivativeY;
double qMag = queryEdgeInforms[cn].Magnitude;
if((txD!=0 || tyD != 0) && (qxD != 0 || qyD != 0))
{
sum += (txD * qxD + tyD * qyD) * qMag / tMag;
}
corSum += 1;
partialScore = sum / corSum;
double curJudge = Math.Min((minScore - 1) + normGreediness * normMinScore, normMinScore * corSum);
if (partialScore < curJudge)
{
break;
}
}
if (partialScore > resultScore)
{
resultScore = partialScore;
conformPoints = new Point(col, row);
}
}
}
//img.Circle(conformPoints, 3, Scalar.Blue);
//img.Rectangle(new Rect(conformPoints, validSize), Scalar.Red);
//ImageShow("hhhh", img);
//二次筛选
if (resultScore > 0.5)
{
if(conformPoints.X+validSize.Width>Width||
conformPoints.Y + validSize.Height > Height)
{
resultScore = 0;
}
}
return resultScore;
}
}
}
}
