1
Contour Features {#tutorial_py_contour_features}
7
In this article, we will learn
9
- To find the different features of contours, like area, perimeter, centroid, bounding box etc
10
- You will see plenty of functions related to contours.
15
Image moments help you to calculate some features like center of mass of the object, area of the
16
object etc. Check out the wikipedia page on [Image
17
Moments](http://en.wikipedia.org/wiki/Image_moment)
19
The function **cv2.moments()** gives a dictionary of all moment values calculated. See below:
24
img = cv2.imread('star.jpg',0)
25
ret,thresh = cv2.threshold(img,127,255,0)
26
contours,hierarchy = cv2.findContours(thresh, 1, 2)
32
From this moments, you can extract useful data like area, centroid etc. Centroid is given by the
33
relations, \f$C_x = \frac{M_{10}}{M_{00}}\f$ and \f$C_y = \frac{M_{01}}{M_{00}}\f$. This can be done as
36
cx = int(M['m10']/M['m00'])
37
cy = int(M['m01']/M['m00'])
43
Contour area is given by the function **cv2.contourArea()** or from moments, **M['m00']**.
45
area = cv2.contourArea(cnt)
51
It is also called arc length. It can be found out using **cv2.arcLength()** function. Second
52
argument specify whether shape is a closed contour (if passed True), or just a curve.
54
perimeter = cv2.arcLength(cnt,True)
57
4. Contour Approximation
58
------------------------
60
It approximates a contour shape to another shape with less number of vertices depending upon the
61
precision we specify. It is an implementation of [Douglas-Peucker
62
algorithm](http://en.wikipedia.org/wiki/Ramer-Douglas-Peucker_algorithm). Check the wikipedia page
63
for algorithm and demonstration.
65
To understand this, suppose you are trying to find a square in an image, but due to some problems in
66
the image, you didn't get a perfect square, but a "bad shape" (As shown in first image below). Now
67
you can use this function to approximate the shape. In this, second argument is called epsilon,
68
which is maximum distance from contour to approximated contour. It is an accuracy parameter. A wise
69
selection of epsilon is needed to get the correct output.
71
epsilon = 0.1*cv2.arcLength(cnt,True)
72
approx = cv2.approxPolyDP(cnt,epsilon,True)
74
Below, in second image, green line shows the approximated curve for epsilon = 10% of arc length.
75
Third image shows the same for epsilon = 1% of the arc length. Third argument specifies whether
76
curve is closed or not.
78
![image](images/approx.jpg)
83
Convex Hull will look similar to contour approximation, but it is not (Both may provide same results
84
in some cases). Here, **cv2.convexHull()** function checks a curve for convexity defects and
85
corrects it. Generally speaking, convex curves are the curves which are always bulged out, or
86
at-least flat. And if it is bulged inside, it is called convexity defects. For example, check the
87
below image of hand. Red line shows the convex hull of hand. The double-sided arrow marks shows the
88
convexity defects, which are the local maximum deviations of hull from contours.
90
![image](images/convexitydefects.jpg)
92
There is a little bit things to discuss about it its syntax:
94
hull = cv2.convexHull(points[, hull[, clockwise[, returnPoints]]
98
- **points** are the contours we pass into.
99
- **hull** is the output, normally we avoid it.
100
- **clockwise** : Orientation flag. If it is True, the output convex hull is oriented clockwise.
101
Otherwise, it is oriented counter-clockwise.
102
- **returnPoints** : By default, True. Then it returns the coordinates of the hull points. If
103
False, it returns the indices of contour points corresponding to the hull points.
105
So to get a convex hull as in above image, following is sufficient:
107
hull = cv2.convexHull(cnt)
109
But if you want to find convexity defects, you need to pass returnPoints = False. To understand it,
110
we will take the rectangle image above. First I found its contour as cnt. Now I found its convex
111
hull with returnPoints = True, I got following values:
112
[[[234 202]], [[ 51 202]], [[ 51 79]], [[234 79]]] which are the four corner points of rectangle.
113
Now if do the same with returnPoints = False, I get following result: [[129],[ 67],[ 0],[142]].
114
These are the indices of corresponding points in contours. For eg, check the first value:
115
cnt[129] = [[234, 202]] which is same as first result (and so on for others).
117
You will see it again when we discuss about convexity defects.
119
6. Checking Convexity
120
---------------------
122
There is a function to check if a curve is convex or not, **cv2.isContourConvex()**. It just return
123
whether True or False. Not a big deal.
125
k = cv2.isContourConvex(cnt)
128
7. Bounding Rectangle
129
---------------------
131
There are two types of bounding rectangles.
133
### 7.a. Straight Bounding Rectangle
135
It is a straight rectangle, it doesn't consider the rotation of the object. So area of the bounding
136
rectangle won't be minimum. It is found by the function **cv2.boundingRect()**.
138
Let (x,y) be the top-left coordinate of the rectangle and (w,h) be its width and height.
140
x,y,w,h = cv2.boundingRect(cnt)
141
cv2.rectangle(img,(x,y),(x+w,y+h),(0,255,0),2)
144
### 7.b. Rotated Rectangle
146
Here, bounding rectangle is drawn with minimum area, so it considers the rotation also. The function
147
used is **cv2.minAreaRect()**. It returns a Box2D structure which contains following detals - (
148
center (x,y), (width, height), angle of rotation ). But to draw this rectangle, we need 4 corners of
149
the rectangle. It is obtained by the function **cv2.boxPoints()**
151
rect = cv2.minAreaRect(cnt)
152
box = cv2.boxPoints(rect)
154
cv2.drawContours(img,[box],0,(0,0,255),2)
156
Both the rectangles are shown in a single image. Green rectangle shows the normal bounding rect. Red
157
rectangle is the rotated rect.
159
![image](images/boundingrect.png)
161
8. Minimum Enclosing Circle
162
---------------------------
164
Next we find the circumcircle of an object using the function **cv2.minEnclosingCircle()**. It is a
165
circle which completely covers the object with minimum area.
167
(x,y),radius = cv2.minEnclosingCircle(cnt)
168
center = (int(x),int(y))
170
cv2.circle(img,center,radius,(0,255,0),2)
172
![image](images/circumcircle.png)
174
9. Fitting an Ellipse
175
---------------------
177
Next one is to fit an ellipse to an object. It returns the rotated rectangle in which the ellipse is
180
ellipse = cv2.fitEllipse(cnt)
181
cv2.ellipse(img,ellipse,(0,255,0),2)
183
![image](images/fitellipse.png)
188
Similarly we can fit a line to a set of points. Below image contains a set of white points. We can
189
approximate a straight line to it.
191
rows,cols = img.shape[:2]
192
[vx,vy,x,y] = cv2.fitLine(cnt, cv2.DIST_L2,0,0.01,0.01)
193
lefty = int((-x*vy/vx) + y)
194
righty = int(((cols-x)*vy/vx)+y)
195
cv2.line(img,(cols-1,righty),(0,lefty),(0,255,0),2)
197
![image](images/fitline.jpg)