Harris corner detection opencv

  Corner point

That is, points with obvious characteristics , Intuitive, , It can be regarded as all kinds of images “ horn ”, For example, the top corner of a house , The square corners of a chessboard, etc , For example, the blue dot in the following figure ：

 

Harris Corner detection

Divide the original image into small windows , take ：

here ,Ix and Iy Namely x and y The image derivative of the direction .( These can be used cv.Sobel() Easy to find ）

And take

among ：

• det(M) = λ1λ2

• trace(M) = λ1 + λ2

• λ1 and λ2 yes M The eigenvalues of the

• When |R| When I was a child , That is to say λ1 and λ2 When they were young , The area is flat .

• When R < 0 when , That is to say λ1 >> λ2 Or vice versa , This area is the edge .

• When R When a large , That is to say λ1 and λ2 It's big and λ1∼λ2 when , This area is a corner .

• dst = cv.cornerHarris(gray,blockSize,ksize,k)
  　　• img -  The data type is  float32  The input image of .
  　　• blockSize -  The domain size to be considered in corner detection .
  　　• ksize - Sobel  Window size used in derivation 
  　　• k - Harris  Free parameters in corner detection equation , The value parameter is  [0,04,0.06].

practice ： 

Original image ：

Code ： 

import cv2.cv2
import numpy as np
import cv2 as cv
from matplotlib import pyplot as plt

img = cv.imread(r'XXXX.jpg')
gray = cv.cvtColor(img,cv.COLOR_BGR2GRAY)
gray = np.float32(gray)
dst = cv.cornerHarris(gray,2,3,0.04)

dst = cv.dilate(dst,None)

img[dst>0.01*dst.max()]=[0,0,255]
cv.imshow('dst',img)
cv.waitKey(0)

result ：