OpenCV : draw, threshold, kernel and edge detection

本文链接：https://blog.youkuaiyun.com/qq_40776179/article/details/105045782

1 install opencv

download opencv_python-4.2.0-cp37-cp37m-win_amd64.whl
cmd: pip install opencv_python-4.2.0-cp37-cp37m-win_amd64.whl

2 Functions

2.1 show a picture

It is used to show a picture.

img1=cv2.imread(1.jpg, cv2.IMREAD_COLOR)
cv2.imshow("img", img1)

ATTENTION:

Don’t use Chinese in the path.
Avoid use special letters as the first letter, such as n, t and so on. Because these letters may be recognized as \n, \t and so on.

2.2 catch a video

cap = cv2.VideoCapture(0)
while(Ture):
    ret, frame= cap.read()
    cv2.imshow("frame", frame)
    cv2.waitKey(0)
cap.release()
cv2.destroyAllWindows()

where:
“ret” is used to show if the camera catch the video;
“frame” is used to store the picture that the camera catch.

2.3 draw a line

img = cv2.line(img, (0,0), (50,40), (255,0,0), 2)

There will be a line begins at (0,0), ends at (50,40), blue and wide 2.

The first parameter is the target picture, the second is the start point, the third is the end point, the fourth is the color(BGR) and the last is the wide.

2.4 Events with mouse

There are many events with mouse, such as left-click, right-click and so on . Using these events, we will be able to do something special with the picture.

def click_event(event, x, y, flags, param):
   if event == cv2.EVENT_LBUTTONDBLCLK:
       print(x, ",", y)
cv2.setMouseCallback("image", click_event)

“setMouseCallback” is used to combine the function with window.

It will print the location where we click the left key.

2.6 Bit-wise operation

bit_and = cv2.bitwise_and(img2, img1)

在这里插入图片描述
Between these 2 images, bit_and will show 0, if there is at least one 0 in img2 and img1.

2.7 Threshold

_, th1 = cv2.threshold(img, 127, 255, cv2.THRESH_BINARY)

在这里插入图片描述

It means 0-127 will become 0 and 127-255 will become 255.

There are some other parameters:

cv2.THRESH_BINARY
cv2.THRESH_BINARY_INV
cv2.THRESH_TRUNC
cv2.THRESH_TOZERO
cv2.THRESH_TOZERO_INV

2.8 Kernel

kernel = np.ones((5, 5), np.float32)/25
dst = cv2. filter2D(img, -1, kernel)
blur = cv2.blur(img, (5,5))
gblur = GaussianBlur(img, (5,5),0)
median = medianBlur(img,5)

在这里插入图片描述

“filter2D” makes the edge smooth.
“blur” filters some noise but the picture may be more vague.
“medianBlur” and “medianBlur” are familiar with “blur”, but they may be more useful.

2.9 Edge detection

lap = cv2. Laplacian(img, cv2.CV_64F, ksize = 3)
lap = np.uint8(np.absolute(lap))

在这里插入图片描述

If we change ksize from 3 to 5:
在这里插入图片描述