//杂质检测
public static List <List <InspectionRect> > ForeignInspection(List <Mat> imgs, List <List <int> > liquilevel, ref List <Mat> retimgs, int inspection_count, int inspection_minarea, int netindex)
{
List <List <InspectionRect> > rectlist = new List <List <InspectionRect> >();
Mat mask = GetTimplateImage("mask.jpg");
Cv2.Resize(mask, mask, new OpenCvSharp.Size(64, 128));
foreach (var item in imgs)
{
PyTorchSDK.AddImage(Unet[netindex], item.Data);
}
PyTorchSDK.Test(Unet[netindex], false);
var data = PyTorchSDK.GetOutPut(Unet[netindex]);
int index = 0;
while (data != IntPtr.Zero)
{
Mat img_tem = new Mat(128, 64, MatType.CV_8U, data);
//要为奇数//膨胀
//Mat se = Cv2.GetStructuringElement(MorphShapes.Rect, new OpenCvSharp.Size(1, 2), new OpenCvSharp.Point(0, 0));
//Cv2.Dilate(img_tem, img_tem, se, new OpenCvSharp.Point(-1, -1), 1);
img_tem = img_tem.Threshold(inspection_count, 255, ThresholdTypes.Binary);
retimgs.Add(img_tem.Clone());
OpenCvSharp.Point[][] contours;
HierarchyIndex[] hierarchyIndexes;
Mat img_masked = new Mat();
img_tem.CopyTo(img_masked, mask);
img_masked.FindContours(out contours, out hierarchyIndexes, RetrievalModes.CComp, ContourApproximationModes.ApproxSimple);
rectlist.Add(new List <InspectionRect>());
for (int i = 0; i < contours.Length; i++)
{
var rect = Cv2.BoundingRect(contours[i]);
if (rect.Width * rect.Height > inspection_minarea && (rect.Y > liquilevel[index][0] || rect.Y + rect.Height > liquilevel[index][0]))
{
rectlist[rectlist.Count - 1].Add(new InspectionRect(rect, rect.Width * rect.Height, rect.Width * rect.Height));
}
}
data = PyTorchSDK.GetOutPut(Unet[netindex]);
index++;
}
return(rectlist);
}
public static string CheckOrc(List <Mat> imgs, int index)
{
string orc_str = "";
if (imgs.Count != 0)
{
foreach (var img in imgs)
{
PyTorchSDK.AddImage(Ocrnet[index], img.Data);
}
PyTorchSDK.Test(Ocrnet[index], true);
var data = PyTorchSDK.GetOutPutInt(Ocrnet[index]);
while (data != -1)
{
orc_str += data;
data = PyTorchSDK.GetOutPutInt(Ocrnet[index]);
}
}
return(orc_str);
}
