public static double CalcSimilarDegree(string filePathOrigin, string filePathTemp)
{
Mat mat1;
Mat mat2;
Mat mat11 = new Mat();
Mat mat22 = new Mat();
Image originImage = Image.FromFile(filePathOrigin);
Image tempImage = Image.FromFile(filePathTemp);
Bitmap bitmap1 = new Bitmap(originImage);
Bitmap bitmap2 = new Bitmap(tempImage);
originImage.Dispose();
tempImage.Dispose();
if (bitmap1.Width > bitmap2.Width && bitmap1.Height > bitmap2.Height)
{
bitmap1 = new Bitmap(ZoomPicture(bitmap1, bitmap2.Width, bitmap2.Height));
}
else
{
bitmap2 = new Bitmap(ZoomPicture(bitmap2, bitmap1.Width, bitmap1.Height));
}
mat1 = bitmap1.ToMat();
mat2 = bitmap2.ToMat();
Cv2.CvtColor(mat1, mat11, ColorConversionCodes.BGR2GRAY);
Cv2.CvtColor(mat2, mat22, ColorConversionCodes.BGR2GRAY);
mat11.ConvertTo(mat11, MatType.CV_32F);
mat22.ConvertTo(mat22, MatType.CV_32F);
double score = Cv2.CompareHist(mat11, mat22, HistCompMethods.Correl) * 100;
return(score);
}
private void CompareHistogramWithStepImage(Mat _removeBackgorundImage)
{
_removeBackgorundImage.ConvertTo(_removeBackgorundImage, MatType.CV_32F);
Cv2.Normalize(_removeBackgorundImage,
openCVImage.stepImage[currentStep]); // , 1.0, 0.0, NormTypes.L1
compareValue = Cv2.CompareHist(openCVImage.stepImage[currentStep],
_removeBackgorundImage, HistCompMethods.Chisqr);
}
static double Compare(string str1, string str2, HistCompMethods method)
{
Mat mat1 = new Mat(str1, ImreadModes.AnyColor);
Mat mat2 = new Mat(@str1, ImreadModes.AnyColor);
Mat mat3 = new Mat();
var value = Cv2.CompareHist(mat1, mat2, method);
Console.WriteLine(new FileInfo(str1).Name + "-" + new FileInfo(str2).Name + ":" + value);
return(value);
}
public double[] CompareHist(Mat img, Mat template, bool showImage = false)
{
Mat[] imgs = HistogramCalculation(img, showImage);
Mat[] templates = HistogramCalculation(template, showImage);
double[] ratios = new double[3];
for (int i = 0; i < imgs.Count(); i++)
{
ratios[i] = Cv2.CompareHist(imgs[i], templates[i], HistCompMethods.Bhattacharyya);
}
return(ratios);
}
private void CompareHistogram(Mat _dstImage, int step)
{
_dstImageBuffer = _dstImage.Clone();
stepImageBufferOfBuffer = stepImageBuffer[step].Clone();
_dstImageBuffer.ConvertTo(_dstImageBuffer, MatType.CV_32F);
stepImageBufferOfBuffer.ConvertTo(stepImageBufferOfBuffer, MatType.CV_32F);
Cv2.Normalize(_dstImageBuffer, stepImageBufferOfBuffer);
openCVImage.stepHist[step] = Cv2.CompareHist(stepImageBufferOfBuffer,
_dstImageBuffer, HistCompMethods.Chisqr);
}
public double CompareImageByHist(Mat img, Mat refImg, bool showImage = false)
{
Mat imgHsv = new Mat();
Mat refImgHsv = new Mat();
Cv2.CvtColor(img, imgHsv, ColorConversionCodes.RGB2HSV);
Cv2.CvtColor(refImg, refImgHsv, ColorConversionCodes.RGB2HSV);
Mat[] imgHsvs = Cv2.Split(imgHsv);
Mat[] refImgHsvs = Cv2.Split(refImgHsv);
int bin1 = 50;
int bin2 = 60;
int[] bins = { bin1, bin2 };
int[] channels = { 0, 1 };
Rangef[] ranges = new Rangef[]
{
new Rangef(0, 180),
new Rangef(0, 256)
};
Mat imgHist = new Mat(img.Size(), MatType.CV_32FC2);
Mat refImgHist = new Mat(img.Size(), MatType.CV_32FC2);
Cv2.CalcHist(imgHsvs, channels, new Mat(), imgHist, 2, bins, ranges, true, false);
Cv2.Normalize(imgHist, imgHist, 1, 0, NormTypes.MinMax, -1, null);
Cv2.CalcHist(refImgHsvs, channels, new Mat(), refImgHist, 2, bins, ranges, true, false);
Cv2.Normalize(refImgHist, refImgHist, 1, 0, NormTypes.MinMax, -1, null);
double ratio = Cv2.CompareHist(imgHist, refImgHist, HistCompMethods.KLDiv);
if (showImage == true)
{
Mat img1 = img.Clone();
Cv2.PutText(img1, ratio.ToString(), new Point(50, 50), HersheyFonts.HersheyPlain, 1, new Scalar(0, 255, 0), 2, LineTypes.AntiAlias);
Cv2.ImShow("CompareHistTestVSRef", img1);
Cv2.WaitKey();
Cv2.DestroyWindow("CompareHistTestVSRef");
}
return(ratio);
}
SquareClassificationScores[,] CalcSquareClassifications(Mat frame)
{
var result = new SquareClassificationScores[8, 8];
Mat squareMask = new Mat(frame.Size(), MatType.CV_8U);
var histSize = new[] { 16, 16, 16 };
var histRanges = new[] { new Rangef(0f, 256f), new Rangef(0f, 256f), new Rangef(0f, 256f) };
for (int row = 0; row < 8; row++)
{
for (int col = 0; col < 8; col++)
{
squareMask.SetTo(Scalar.Black);
_histChessboardModel.DrawSquareMask(row, col, squareMask, HistChessboardModel.SquareMaskScale);
Mat squareHist = new Mat();
Cv2.CalcHist(
images: new[] { frame },
channels: new int[] { 0, 1, 2 },
mask: squareMask,
hist: squareHist,
dims: 3,
histSize: histSize,
ranges: histRanges);
var method = HistCompMethods.Correl;
var squareColor = HistChessboardModel.SquareColor(row, col);
var likeWhitePiece = Cv2.CompareHist(
h1: squareHist,
h2: _histChessboardModel.XOnYHist[(int)ChessColor.White, squareColor],
method: method);
var likeBlackPiece = Cv2.CompareHist(
h1: squareHist,
h2: _histChessboardModel.XOnYHist[(int)ChessColor.Black, squareColor],
method: method);
var likeEmpty = Cv2.CompareHist(
h1: squareHist,
h2: _histChessboardModel.XOnYHist[(int)SquareState.Empty, squareColor],
method: method);
result[row, col] = new SquareClassificationScores(new [] { likeWhitePiece, likeBlackPiece, likeEmpty });
}
}
return(result);
}
/// <summary>
/// 通过直方图比较
/// </summary>
/// <param name="sourceImage"></param>
/// <param name="targetImage"></param>
/// <param name="method"></param>
/// <returns></returns>
public static double CompareImageByHistogram(string sourceImage, string targetImage, HistCompMethods method = HistCompMethods.KLDiv)
{
using (Mat source = new Mat(sourceImage, ImreadModes.AnyColor | ImreadModes.AnyDepth))
using (Mat target = new Mat(targetImage, ImreadModes.AnyColor | ImreadModes.AnyDepth))
{
//1:从BGR空间转换到HSV色彩空间
Mat src1HSV = new Mat();
Mat src2HSV = new Mat();
Cv2.CvtColor(source, src1HSV, ColorConversionCodes.BGR2HSV);
Cv2.CvtColor(target, src2HSV, ColorConversionCodes.BGR2HSV);
//Mat[] mats = new Mat[] { baseHSV, src1HSV, src2HSV };
Mat[] mats2 = Cv2.Split(src1HSV);
Mat[] mats3 = Cv2.Split(src2HSV);
//计算直方图并归一化数据
int bin1 = 50; //灰度等级
int bin2 = 60; //灰度等级
int[] histSiz = { bin1, bin2 };
int[] channels = { 0, 1 }; //图像通道数
Rangef[] rangefs = new Rangef[] //梯度值范围
{
new Rangef(0, 180),
new Rangef(0, 256)
};
Mat Img2 = new Mat(source.Size(), MatType.CV_32FC2);
Mat Img3 = new Mat(target.Size(), MatType.CV_32FC2);
Cv2.CalcHist(mats2, channels, new Mat(), Img2, 2, histSiz, rangefs, true, false);
Cv2.Normalize(Img2, Img2, 0, 1, NormTypes.MinMax, -1, null);
Cv2.CalcHist(mats3, channels, new Mat(), Img3, 2, histSiz, rangefs, true, false);
Cv2.Normalize(Img3, Img3, 0, 1, NormTypes.MinMax, -1, null);
//比较直方图
double s1coms2 = Cv2.CompareHist(Img2, Img3, method);
return(s1coms2);
}
}
public void matchTestCompareHist()
{
Mat img = new Mat(Images.picTest, ImreadModes.Color);
Mat refImg = new Mat(Images.picSetRef, ImreadModes.Color);
Mat imgHsv = new Mat();
Mat refImgHsv = new Mat();
Cv2.CvtColor(img, imgHsv, ColorConversionCodes.RGB2HSV);
Cv2.CvtColor(refImg, refImgHsv, ColorConversionCodes.RGB2HSV);
Mat[] imgHsvs = Cv2.Split(imgHsv);
Mat[] refImgHsvs = Cv2.Split(refImgHsv);
int bin1 = 50;
int bin2 = 60;
int[] bins = { bin1, bin2 };
int[] channels = { 0, 1 };
Rangef[] ranges = new Rangef[]
{
new Rangef(0, 180),
new Rangef(0, 256)
};
Mat imgHist = new Mat(img.Size(), MatType.CV_32FC2);
Mat refImgHist = new Mat(img.Size(), MatType.CV_32FC2);
Cv2.CalcHist(imgHsvs, channels, new Mat(), imgHist, 2, bins, ranges, true, false);
Cv2.Normalize(imgHist, imgHist, 1, 0, NormTypes.MinMax, -1, null);
Cv2.CalcHist(refImgHsvs, channels, new Mat(), refImgHist, 2, bins, ranges, true, false);
Cv2.Normalize(refImgHist, refImgHist, 1, 0, NormTypes.MinMax, -1, null);
double ratio = Cv2.CompareHist(refImgHist, imgHist, HistCompMethods.Bhattacharyya);
}
public decimal Compare(Mat first, Mat second)
{
var result = Cv2.CompareHist(first, second, HistCompMethods.Chisqr);
return((decimal)Math.Round(result, 2));
}
private static void HistogarmToComparison(string path1, string path2, string path3)
{
using (Mat baseMat = new Mat(path1, ImreadModes.AnyColor | ImreadModes.AnyDepth))
using (Mat src1 = new Mat(path2, ImreadModes.AnyColor | ImreadModes.AnyDepth))
using (Mat src2 = new Mat(path3, ImreadModes.AnyColor | ImreadModes.AnyDepth))
{
//1:从BGR空间转换到HSV色彩空间
Mat baseHSV = new Mat();
Mat src1HSV = new Mat();
Mat src2HSV = new Mat();
Cv2.CvtColor(baseMat, baseHSV, ColorConversionCodes.BGR2HSV);
Cv2.CvtColor(src1, src1HSV, ColorConversionCodes.BGR2HSV);
Cv2.CvtColor(src2, src2HSV, ColorConversionCodes.BGR2HSV);
//Mat[] mats = new Mat[] { baseHSV, src1HSV, src2HSV };
Mat[] mats1 = Cv2.Split(baseHSV);
Mat[] mats2 = Cv2.Split(src1HSV);
Mat[] mats3 = Cv2.Split(src2HSV);
//计算直方图并归一化数据
int bin1 = 50; //灰度等级
int bin2 = 60; //灰度等级
int[] histSiz = { bin1, bin2 };
int[] channels = { 0, 1 }; //图像通道数
Rangef[] rangefs = new Rangef[] //梯度值范围
{
new Rangef(0, 180),
new Rangef(0, 256)
};
Mat Img1 = new Mat(baseMat.Size(), MatType.CV_32FC2);
Mat Img2 = new Mat(src1.Size(), MatType.CV_32FC2);
Mat Img3 = new Mat(src2.Size(), MatType.CV_32FC2);
HistCompMethods method = HistCompMethods.KLDiv;
Cv2.CalcHist(mats1, channels, new Mat(), Img1, 2, histSiz, rangefs, true, false);
Cv2.Normalize(Img1, Img1, 0, 1, NormTypes.MinMax, -1, null);
Cv2.CalcHist(mats2, channels, new Mat(), Img2, 2, histSiz, rangefs, true, false);
Cv2.Normalize(Img2, Img2, 0, 1, NormTypes.MinMax, -1, null);
Cv2.CalcHist(mats3, channels, new Mat(), Img3, 2, histSiz, rangefs, true, false);
Cv2.Normalize(Img3, Img3, 0, 1, NormTypes.MinMax, -1, null);
//比较直方图
double bcomb = Cv2.CompareHist(Img1, Img1, method);
double s1coms2 = Cv2.CompareHist(Img2, Img3, method);
double bcoms1 = Cv2.CompareHist(Img1, Img2, method);
double bcoms2 = Cv2.CompareHist(Img1, Img3, method);
Mat mats1Ands2 = new Mat();
src2.CopyTo(mats1Ands2);
Cv2.PutText(baseMat, bcomb.ToString(), new Point(30, 30), HersheyFonts.HersheyComplex, 1, new Scalar(0, 255, 0), 2, LineTypes.AntiAlias);
Cv2.PutText(src1, bcoms1.ToString(), new Point(30, 30), HersheyFonts.HersheyComplex, 1, new Scalar(0, 255, 0), 2, LineTypes.AntiAlias);
Cv2.PutText(src2, bcoms2.ToString(), new Point(30, 30), HersheyFonts.HersheyComplex, 1, new Scalar(0, 255, 0), 2, LineTypes.AntiAlias);
Cv2.PutText(mats1Ands2, s1coms2.ToString(), new Point(30, 30), HersheyFonts.HersheyComplex, 1, new Scalar(0, 255, 0), 2, LineTypes.AntiAlias);
using (new Window("Img1 :Img1", WindowMode.Normal, baseMat))
using (new Window("Img1 :Img2", WindowMode.Normal, src1))
using (new Window("Img1 :Img3", WindowMode.Normal, src2))
{
Cv2.WaitKey(0);
}
}
}
public static bool IsUnique(Mat face)
{
if (_faces.Count == 0)
{
var color1 = ColorsExtractor.ExtractMiddleColor(face);
color1 = ColorsExtractor.ClosestColorRgb(color1);
_faces.Add(face);
_facesMiddleColors.Add(color1);
return(true);
}
/// Using 50 bins for hue and 60 for saturation
int h_bins = 50;
int s_bins = 60;
int[] histSize = new int[2] {
h_bins, s_bins
};
// hue varies from 0 to 179, saturation from 0 to 255
var h_ranges = new Rangef(0, 180);
var s_ranges = new Rangef(0, 256);
var ranges = new Rangef[] { h_ranges, s_ranges };
// Use the o-th and 1-st channels
int[] channels = new int[2] {
0, 1
};
var faceHsv = face.CvtColor(ColorConversionCodes.BGR2HSV);
var faceHistogram = new Mat();
Cv2.CalcHist(new Mat[] { faceHsv }, channels, new Mat(), faceHistogram, 2, histSize, ranges, true, false);
Cv2.Normalize(faceHistogram, faceHistogram, 0, 1, NormTypes.MinMax, -1, new Mat());
var color = ColorsExtractor.ExtractMiddleColor(face);
foreach (var f in _faces)
{
var fHsv = f.CvtColor(ColorConversionCodes.BGR2HSV);
var fHistogram = new Mat();
Cv2.CalcHist(new Mat[] { fHsv }, channels, new Mat(), fHistogram, 2, histSize, ranges, true, false);
Cv2.Normalize(fHistogram, fHistogram, 0, 1, NormTypes.MinMax, -1, new Mat());
double correl = Cv2.CompareHist(faceHistogram, fHistogram, HistCompMethods.Correl);
//if (correl > 0.5) return false;
bool contains = _facesMiddleColors.Contains(color);
if (correl > 0.4 && contains)
{
return(false);
}
//if (correl > 0.65) return false;
}
color = ColorsExtractor.ExtractMiddleColor(face);
_faces.Add(face);
_facesMiddleColors.Add(color);
return(true);
}
private void Analyze()
{
ArrayList position = CountImage();
int length = position.Count;
Mat[] srcImage = new Mat[length];
Mat[] hsvImage = new Mat[length];
Mat[] histogram = new Mat[length];
Mat originSrcImage = new Mat(mOriginImageURL);
Mat originHsvImage = new Mat();
Mat originHistogram = new Mat();
for (int i = 0; i < length; i++)
{
srcImage[i] = new Mat(mSimilarImage[(int)position[i]].URL);
hsvImage[i] = new Mat();
Cv2.CvtColor(srcImage[i], hsvImage[i], ColorConversionCodes.BGR2HSV);
}
Cv2.CvtColor(originSrcImage, originHsvImage, ColorConversionCodes.BGR2HSV);
int hBins = 50;
int sBins = 60;
int[] histSize = { hBins, sBins };
Rangef[] ranges = { new Rangef(0, 180), new Rangef(0, 256) };
int[] channels = { 0, 1 };
for (int i = 0; i < length; i++)
{
histogram[i] = new Mat();
Cv2.CalcHist(
images: new[] { hsvImage[i] },
channels: channels,
mask: null,
hist: histogram[i],
dims: 1,
histSize: histSize,
ranges: ranges);
Cv2.Normalize(histogram[i], histogram[i], 0, 1, NormTypes.MinMax);
}
Cv2.CalcHist(
images: new[] { originHsvImage },
channels: channels,
mask: null,
hist: originHistogram,
dims: 1,
histSize: histSize,
ranges: ranges);
Cv2.Normalize(originHistogram, originHistogram, 0, 1, NormTypes.MinMax);
double maxValue = 0;
int maxID = 0;
for (int i = 0; i < length; i++)
{
double bass = Cv2.CompareHist(originHistogram, histogram[i], HistCompMethods.Correl);
if (bass < 0)
{
bass = 0;
}
if (i == 0)
{
maxValue = bass;
maxID = 0;
}
else
{
if (maxValue < bass)
{
maxValue = bass;
maxID = i;
}
}
mSimilarity[(int)position[i]] = (float)bass * 100;
MyLog.LogManager.Log("SimilarImage-Analyze-Method[" + (int)position[i] + "] Perfect: " + mSimilarity[(int)position[i]]);
}
MyLog.LogManager.Log("SimilarImage-Analyze-가장 유사한 그림의 유사도: " + maxValue);
Dispatcher.Invoke(new Action(delegate()
{
BitmapImage tmpImage = new BitmapImage();
tmpImage.BeginInit();
tmpImage.CacheOption = BitmapCacheOption.OnLoad;
mResultIndex = (int)position[maxID];
tmpImage.UriSource = new Uri(mSimilarImage[mResultIndex].URL);
tmpImage.EndInit();
similar_result_image.Source = tmpImage;
similar_analyze.Content = FindResource(Constants.STRING_BUTTON_COMPLETE);
similar_analyze.IsEnabled = false;
similar_btn_target_histogram.IsEnabled = true;
similar_btn_result_histogram.IsEnabled = true;
StringBuilder str = new StringBuilder();
for (int i = 0; i < length; i++)
{
str.Clear();
str.Append((string)FindResource(Constants.STRING_SIMILAR_LABEL));
str.Append(string.Format(" {0:0.####}", mSimilarity[(int)position[i]]));
str.Append("%");
mSimilarLabel[(int)position[i]].Content = str.ToString();
}
}));
}
