Contour Features Boundary characteristics

Find Contour findcontours

1 cv2.findcontours (image, Mode, method[, contours[, hierarchy[, offset]]) →image, contours, hierarchy

Parameter interpretation

• Image: Original images, can be obtained by compare (), InRange (), Threshold () and other binary image image
• Mode: Contour Retrieval modes
• Method: Contour Approximation methods

The mode parameter has the desired value

1. Cv_retr_external retrieves only external outlines.
2. Cv_retr_list retrieves all outlines but does not establish hierarchical relationships.
3. Cv_retr_ccomp retrieves all the outlines and establishes a level two hierarchy.
4. Cv_retr_tree retrieves all the outlines and establishes a nested outline hierarchy.

The method parameter is preferable

1. Cv_chain_approx_none stores all contour points.
2. Cv_chain_approx_simple compresses horizontal, vertical, and diagonal lines, leaving only their endpoints.
3. Cv_chain_approx_tc89_l1,cv_chain_approx_tc89_kcos applied a style of the teh-chin-chain approximation algorithm

Calculate the contour area of an image Cv2.contourarea ()

1 Cv2.contourarea (contour[, oriented]) →retval

calculate the moment of the image cv2.moments ()

1 cv2.moments (array[, binaryimage]) →retval

See code for details

1 ImportCv22 ImportNumPy as NP3  4img = Cv2.imread ('star.jpg', 0)5Ret,thresh = Cv2.threshold (img,127,255, 0)6Contours,hierarchy =cv2.findcontours (Thresh, CV. Retr_external, CV. Chain_approx_simple)7  8CNT =Contours[0]9M =cv2.moments (CNT)Ten Print(Cv2.contourarea (CNT)) One Print(M) A   - #get {' m00 ': 0.5, ' M10 ': 53.83333333333333, ' m01 ': 367.3333333333333, ' M20 ': 5796.083333333333, ' M11 ': 39549.541666666664, ' m02 ': 269867.5833333333, ' M30 ': 624050.9500000001, ' M21 ': 4258186.233333333, ' M12 ': 29055722.733333334, ' m03 ': 198262758.70000002, ' mu20 ': 0.027777777778283053, ' mu11 ': -0.01388888888322981, ' mu02 ': 0.027777777810115367, ' mu30 ': -0.003703703638166189, ' mu21 ': 0.0018518519221615293, ' mu12 ': 0.001851847569924292, ' Mu03 ': -0.0037037134170532227, ' nu20 ': 0.11111111111313221, ' nu11 ': -0.05555555553291924, ' nu02 ': 0.11111111124046147 , ' nu30 ': -0.020951311664420796, ' nu21 ': 0.01047565641531008, ' nu12 ': 0.010475631795338369, ' nu03 ':- 0.020951366982159467} -   the #we use this to get the center of gravity . -CX = Int (m['M10']/m['m00']) -cy = Int (m['M01']/m['m00']) - #the results of Contourarea are consistent with the results of m00. +  

Contour Circumference Calculation cv2.arclength ()

1 perimeter = cv2.arclength (cnt,true)

• The first parameter is a contour
• The second parameter specifies whether the shape is a closed contour, true is closed, otherwise the curve

Contour Approximation

1 cv2.approxpolydp (curve, Epsilon, closed[, Approxcurve]) →approxcurve

Parameter interpretation

• Curve: Input 2D points, such as Findcontours obtained contour
• Epsilon: Accuracy
• Closed: Is it closed, as it was said before?

The output is approximate after the contour Contour Line Fitting

1 cv2.polylines (IMG, pts, isClosed, color[, thickness[, linetype[, Shift]]) →img

Parameter interpretation

• IMG: Input Image
• PTS: A collection of outlines to fit, for example [contours[1],contours[2]]
• IsClosed: Is it closed, as it was said before?
• Color: Colors, for example (0,0,255)
• Thickness: thickness, 1 2 3, etc.
• Linetype: Type of line
• Shift: Number of decimal digits in fixed-point coordinates

1 ImportCv2 as CV2 ImportNumPy as NP3img = Cv.imread ("test.jpg", 0)4_,contours,_ = Cv.findcontours (img,2,1)5CNT =Contours[0]6Epsilon = 0.01 * Cv.arclength (cnt,true)#here, use arclength to get contour circumference or curve length .7approx =CV.APPROXPOLYDP (cnt,epsilon,true)8out_img = Cv.polylines (Img,[approx],true, (0,0,255), 2)9Cv.imshow ("Image", out_img)TenK = Cv.waitkey (1) & 0xFF One ifk== 27: ACv.destroyallwindows ()

Convex Packet Inspection

1 cv2.convexhull (points[, hull[, clockwise[, Returnpoints]]) →hull

parameter Interpretation

• Points: Input 2D point set, such as findcontours obtained contour
• Hull: Output Convex package
• clockwise: If true, the output convex hull order is clockwise, otherwise counterclockwise

The function returns a convex hull (point set)

1 ImportCv2 as CV2 ImportNumPy as NP3img = Cv.imread ("test.jpg", 0)4_,contours,_ = Cv.findcontours (img,2,1)5CNT = contours[1]6Hull =cv.convexhull (CNT)7out_img = Cv.polylines (Img,[hull],true, (0,255,255), 2)8Cv.imshow ("Image", out_img)9Cv.waitkey (0)

Bounding Rectangle get the straight bounding rectangle

1 cv2.boundingrect (points) →retval

parameter Interpretation

• Points: The set of points given to determine the boundary, such as contour

The function returns the four vertex coordinates of the resulting bounding rectangle get the rotated rectangle

1 cv2.minarearect (points) →retval

Parameter description:

• Points: Findcountours got the Contour

Use

Import Cv2 as CV Import  = Cv.imread ("test.jpg"== = = Cv.minarearect (CNT)# here is the rotation rectangle box = cv.cv.BoxPoints (rect)# Gets the endpoint box = Np.int0 (Box)# rounding down

Minimum closed Circle  

1 cv2.minenclosingcircle (points) →center, radius

 Parameter interpretation

• Points: Input point set, such as contour

Output to Circle center point coordinates and RADIUS   Ellipse Fitting  

1 cv2.fitellipse (points) →retval

  parameter Interpretation

• Points: Input point set, such as contour

 The output is an ellipse whose properties have a center point coordinate, a two-axis length, and a deflection angle   Use

1 ImportCv2 as CV2 ImportNumPy as NP3img = Cv.imread ("test.jpg", 0)4_,contours,_ =cv.findcontours (IMG,CV. RETR_LIST,CV. Chain_approx_simple)5CNT =Contours[0]6Circle =cv.minenclosingcircle (CNT)7Ellipse =cv.fitellipse (CNT)8out_1 = Cv.circle (Img,circle, (0,255,255), 2)9Out_2 =cv.ellipse (Img,ellipse, (0,255,255), 2)TenCv.imshow ("IMG1", out_1) OneCv.imshow ("Img2", Out_2) ACv.waitkey (0)

Contour Features Boundary characteristics