public void ProcessImage(Image <Gray, byte> grayFrame)
{
if (equalizeHist)
{
grayFrame._EqualizeHist();//autocontrast
}
//smoothed
Image <Gray, byte> smoothedGrayFrame = grayFrame.PyrDown();
smoothedGrayFrame = smoothedGrayFrame.PyrUp();
//canny
Image <Gray, byte> cannyFrame = null;
if (noiseFilter)
{
cannyFrame = smoothedGrayFrame.Canny(new Gray(cannyThreshold), new Gray(cannyThreshold));
}
//smoothing
if (blur)
{
grayFrame = smoothedGrayFrame;
}
//binarize
CvInvoke.cvAdaptiveThreshold(grayFrame, grayFrame, 255, Emgu.CV.CvEnum.ADAPTIVE_THRESHOLD_TYPE.CV_ADAPTIVE_THRESH_MEAN_C, Emgu.CV.CvEnum.THRESH.CV_THRESH_BINARY, adaptiveThresholdBlockSize + adaptiveThresholdBlockSize % 2 + 1, adaptiveThresholdParameter);
//
grayFrame._Not();
//
if (addCanny)
{
if (cannyFrame != null)
{
grayFrame._Or(cannyFrame);
}
}
//
this.binarizedFrame = grayFrame;
//dilate canny contours for filtering
if (cannyFrame != null)
{
cannyFrame = cannyFrame.Dilate(3);
}
//find contours
var sourceContours = grayFrame.FindContours(Emgu.CV.CvEnum.CHAIN_APPROX_METHOD.CV_CHAIN_APPROX_NONE, Emgu.CV.CvEnum.RETR_TYPE.CV_RETR_LIST);
//filter contours
contours = FilterContours(sourceContours, cannyFrame, grayFrame.Width, grayFrame.Height);
//find templates
lock (foundTemplates)
foundTemplates.Clear();
samples.Clear();
lock (templates)
Parallel.ForEach <Contour <Point> >(contours, (contour) =>
{
var arr = contour.ToArray();
Template sample = new Template(arr, contour.Area, samples.templateSize);
lock (samples)
samples.Add(sample);
if (!onlyFindContours)
{
FoundTemplateDesc desc = finder.FindTemplate(templates, sample);
if (desc != null)
{
lock (foundTemplates)
foundTemplates.Add(desc);
}
}
}
);
//
FilterByIntersection(ref foundTemplates);
}
public void ProcessImage(Image <Gray, byte> grayFrame)
{
if (equalizeHist)
{
grayFrame._EqualizeHist();//autocontrast
}
//高斯平滑
Image <Gray, byte> smoothedGrayFrame = grayFrame.PyrDown();
smoothedGrayFrame = smoothedGrayFrame.PyrUp();
//canny
Image <Gray, byte> cannyFrame = null;
if (noiseFilter)
{
cannyFrame = smoothedGrayFrame.Canny(cannyThreshold, cannyThreshold);
}
//smoothing
if (blur)
{
grayFrame = smoothedGrayFrame;
}
//局部自适应阈值二值化，阈值本身作为了一个变量，检测更有效
//CvInvoke.cvAdaptiveThreshold(grayFrame, grayFrame, 255, Emgu.CV.CvEnum.ADAPTIVE_THRESHOLD_TYPE.CV_ADAPTIVE_THRESH_MEAN_C, Emgu.CV.CvEnum.THRESH.CV_THRESH_BINARY, adaptiveThresholdBlockSize + adaptiveThresholdBlockSize % 2 + 1, adaptiveThresholdParameter);
CvInvoke.AdaptiveThreshold(grayFrame, grayFrame, 255, Emgu.CV.CvEnum.AdaptiveThresholdType.MeanC, Emgu.CV.CvEnum.ThresholdType.Binary, adaptiveThresholdBlockSize + adaptiveThresholdBlockSize % 2 + 1, adaptiveThresholdParameter);
//
grayFrame._Not();
//
if (addCanny)
{
if (cannyFrame != null)
{
grayFrame._Or(cannyFrame); //试验了一下，这样轮廓会更加明显
}
}
//
this.binarizedFrame = grayFrame;
//dilate canny contours for filtering
//膨胀操作，使白色区域的像素增加一圈，看起来轮廓更细了
if (cannyFrame != null)
{
cannyFrame = cannyFrame.Dilate(3);
}
//find contours
VectorOfVectorOfPoint sourceContours = new VectorOfVectorOfPoint();
CvInvoke.FindContours(grayFrame, sourceContours, null, Emgu.CV.CvEnum.RetrType.List, Emgu.CV.CvEnum.ChainApproxMethod.ChainApproxNone);
//var sourceContours = grayFrame.FindContours(Emgu.CV.CvEnum.CHAIN_APPROX_METHOD.CV_CHAIN_APPROX_NONE, Emgu.CV.CvEnum.RETR_TYPE.CV_RETR_LIST);
//filter contours
contours = FilterContours(sourceContours, cannyFrame, grayFrame.Width, grayFrame.Height);
//find templates
lock (foundTemplates)
foundTemplates.Clear();
samples.Clear();
lock (templates) //多线程，异步执行
Parallel.ForEach <VectorOfPoint>(contours, (contour) =>
{
var arr = contour.ToArray();
Template sample = new Template(arr, CvInvoke.ContourArea(contour), samples.templateSize);
lock (samples)
samples.Add(sample);
if (!onlyFindContours)
{
FoundTemplateDesc desc = finder.FindTemplate(templates, sample);
if (desc != null)
{
lock (foundTemplates)
foundTemplates.Add(desc);
}
}
}
);
//
FilterByIntersection(ref foundTemplates);
}
private void DrawAugmentedReality(FoundTemplateDesc found, Graphics gr)
{
string fileName = Path.GetDirectoryName(templateFile) + "\\" + found.template.name;
if (!AugmentedRealityImages.ContainsKey(fileName))
{
if (!File.Exists(fileName)) return;
AugmentedRealityImages[fileName] = Image.FromFile(fileName);
}
Image img = AugmentedRealityImages[fileName];
Point p = found.sample.contour.SourceBoundingRect.Center();
var state = gr.Save();
gr.TranslateTransform(p.X, p.Y);
MessageBox.Show((180f * found.angle / Math.PI).ToString());
gr.RotateTransform((float)(180f * found.angle / Math.PI));
gr.ScaleTransform((float)(found.scale), (float)(found.scale));
gr.DrawImage(img, new Point(-img.Width/2, -img.Height/2));
gr.Restore(state);
}
public FoundTemplateDesc FindTemplate(Templates templates, Template sample)
{
FoundTemplateDesc desc2;
double num = 0.0;
double angle = 0.0;
Complex complex = new Complex();
Template objA = null;
using (List <Template> .Enumerator enumerator = templates.GetEnumerator())
{
while (true)
{
if (!enumerator.MoveNext())
{
break;
}
Template current = enumerator.Current;
if ((Math.Abs((int)(sample.autoCorrDescriptor1 - current.autoCorrDescriptor1)) <= this.maxACFDescriptorDeviation) && ((Math.Abs((int)(sample.autoCorrDescriptor2 - current.autoCorrDescriptor2)) <= this.maxACFDescriptorDeviation) && ((Math.Abs((int)(sample.autoCorrDescriptor3 - current.autoCorrDescriptor3)) <= this.maxACFDescriptorDeviation) && (Math.Abs((int)(sample.autoCorrDescriptor4 - current.autoCorrDescriptor4)) <= this.maxACFDescriptorDeviation))))
{
double num3 = 0.0;
if (!this.checkACF || (current.autoCorr.NormDot(sample.autoCorr).Norma >= this.minACF))
{
if (this.checkICF)
{
complex = current.contour.InterCorrelation(sample.contour).FindMaxNorma();
num3 = complex.Norma / (current.contourNorma * sample.contourNorma);
if (num3 < this.minICF)
{
continue;
}
if (Math.Abs(complex.Angle) > this.maxRotateAngle)
{
continue;
}
}
if ((!current.preferredAngleNoMore90 || (Math.Abs(complex.Angle) < 1.5707963267948966)) && (num3 >= num))
{
num = num3;
objA = current;
angle = complex.Angle;
}
}
}
}
}
if ((objA != null) && (objA.name == this.antiPatternName))
{
objA = null;
}
if (ReferenceEquals(objA, null))
{
desc2 = null;
}
else
{
desc2 = new FoundTemplateDesc {
template = objA,
rate = num,
sample = sample,
angle = angle
};
}
return(desc2);
}
public void ProcessImage(Image <Gray, byte> grayFrame)
{
Action <Contour <Point> > body = null;
if (this.equalizeHist)
{
grayFrame._EqualizeHist();
}
Image <Gray, byte> image = grayFrame.PyrDown().PyrUp();
Image <Gray, byte> image2 = null;
if (this.noiseFilter)
{
image2 = image.Canny(new Gray((double)this.cannyThreshold), new Gray((double)this.cannyThreshold));
}
if (this.blur)
{
grayFrame = image;
}
CvInvoke.cvAdaptiveThreshold((IntPtr)grayFrame, (IntPtr)grayFrame, 255.0, ADAPTIVE_THRESHOLD_TYPE.CV_ADAPTIVE_THRESH_MEAN_C, THRESH.CV_THRESH_BINARY, (this.adaptiveThresholdBlockSize + (this.adaptiveThresholdBlockSize % 2)) + 1, 1.0);
grayFrame._Not();
if (this.addCanny && (image2 != null))
{
grayFrame._Or(image2);
}
this.binarizedFrame = grayFrame;
if (image2 != null)
{
image2 = image2.Dilate(3);
}
Contour <Point> contours = grayFrame.FindContours(CHAIN_APPROX_METHOD.CV_CHAIN_APPROX_NONE, RETR_TYPE.CV_RETR_LIST);
this.contours = this.FilterContours(contours, image2, grayFrame.Width, grayFrame.Height);
lock (this.foundTemplates)
{
this.foundTemplates.Clear();
}
this.samples.Clear();
lock (this.templates)
{
if (body == null)
{
body = delegate(Contour <Point> contour) {
Template item = new Template(contour.ToArray(), contour.Area, this.samples.templateSize);
lock (this.samples)
{
this.samples.Add(item);
}
if (!this.onlyFindContours)
{
FoundTemplateDesc desc = this.finder.FindTemplate(this.templates, item);
if (desc != null)
{
lock (this.foundTemplates)
{
this.foundTemplates.Add(desc);
}
}
}
};
}
Parallel.ForEach <Contour <Point> >(this.contours, body);
}
FilterByIntersection(ref this.foundTemplates);
}
