You want to do a circle's diameter and center detection, has been canny edge detection and conduct a closed operation (before edge detection filter and gray), and now want to accurately obtain the diameter of the two circles as well as the center of the circle and the output as a result, the oneself first, hope everybody can give some ideas and solutions, thank you

CodePudding user response:

Fyi:

//////////////////////////////////////////////////////////////////////////
//contour is a circle, judged by the area of the contour PI * r * r difference, distinguish the circle, 
//contour: input profile 
//minRadius: radius length at least more than minRadius 
//////////////////////////////////////////////////////////////////////////
BOOL isCircleContour (vector & Contour, double minRadius) 
{
BOOL isCircle=FALSE; 

//double generating=arcLength (contour, TRUE); 
Double generating=arcLength (contour, FALSE); 
Double perimeter2=arcLength (contour, TRUE); 
The Rect the Rect=boundingRect (contour); 
Double r=(. The rect. The rect width + height)/4.0;//external torque radius approximately equal to the contour length/2 

Double area=contourArea (contour); 

Double PI=3.1415926; 

//cout<& lt;" X/y/width/height \ t "& lt; //cout<& lt;" Generating: FALSE \ t "& lt; //cout<& lt;" Perimeter2: TRUE \ t "& lt; //cout<& lt;" Area: \ t "& lt; //cout<& lt;" R: "& lt; 
Double perimeterDiffRatio=fabs (2 * PI * r - generating) * 2.0/(2 * PI * r + generating) * 100; 
Double areaDiffRatio=fabs (PI * r * r - area) * 2.0/(PI * r * r + area) * 100; 
Double sideDiffRatio=abs (the rect. Width - the rect. Height) * 2.0/(. The rect. The rect width + height) * 100; 

//cstrings CstrInfo3; 
//CstrInfo3. The Format (_T (" \ n areaDiffRatio perimeterDiffRatio: % f, % f \ n sideDiffRatio: % f "), perimeterDiffRatio, areaDiffRatio, sideDiffRatio); 
//AfxMessageBox (CstrInfo3); 

If (r> MinRadius) 
{
Cout<& lt;" Perimeter percentage difference: "& lt; Cout<& lt;" Area percentage difference: "& lt; Cout<& lt;" Side length percentage difference: "& lt; Cout} 

//the original 
//const double areaDiffThresh=18.0;//the current statistics, the largest area of poor 2.58% 
//const double perimeterDiffThresh=15;//the current statistics, the largest circumference difference 5.65% 
//const double sideDiffThresh=15;//the current statistics, the largest length difference 7.3% 
//const double avgDiffThresh=15.0;//three difference ratio of weighted average of the request is less than 5.0; 

Const double areaDiffThresh=5;//the current statistics, the largest area of poor 2.58% 
Const double perimeterDiffThresh=7.25;//the current statistics, the largest circumference difference 5.65% 
Const double sideDiffThresh=10;//the current statistics, the largest length difference 7.3% 
Const double avgDiffThresh=5.0;//three difference ratio of weighted average of the request is less than 5.0; 

Double avdDiffRatio=(areaDiffRatio + + sideDiffRatio perimeterDiffRatio)/3.0; 

//calculate the radius of the outline of the perimeter is multiples of 2 * PI 
If ((areaDiffRatio & lt; AreaDiffThresh) 
& & (perimeterDiffRatio & lt; PerimeterDiffThresh) 
& & (sideDiffRatio & lt; SideDiffThresh) 
& & (avdDiffRatio& & (r> MinRadius)) 
{
IsCircle=TRUE; 
} 

Return isCircle; 

} 

CodePudding user response:

Miss zhao, I this two circles can be detected? If in the process of filming is not vertical, will get round a little oval effect how to eliminate the influence thank you teacher

CodePudding user response:

Relaxation limit scope of bai

CodePudding user response:

Hough transform. There are code! Don't have to use opencv

CodePudding user response:

The Matlab solution is more convenient, binarization, and then use regionprops the function, find out the local area, center and other properties