Image feature Harris corner detection

 ( Generally speaking, corner points are along x Direction 、 Along the y The change of direction is obvious ; The boundary changes only in one direction ）

The basic principle

（ Equivalent to the change of gray value ）

 

 

Based on Taylor expansion , To image I(x,y) In translation （xi,yi) Then we make the first order approximation ：

 

 

 

 

 opencv- Corner detection effect

cv2.cornerHarris(img,blockSize,ksize,k)

1、 img: The data type is float32 The input image of

2、blockSize: The size of the specified area in corner detection

3、ksize:sobel Window size used in derivation （ commonly ksize=3)

4、k: retrieval argument 【0.04,0.06】

Code ：

import cv2
import numpy as np
 
img=cv2.imread('E:/opencv/chessboard.jpg')
print('img.shape:',img.shape)
gray=cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
#gray=np.float32(gray)
dst=cv2.cornerHarris(gray,2,3,0.04)  # Self similarity 
print('dst.shape:',dst.shape)
 
img[dst>0.01*dst.max()]=[0,0,255]   # As long as the position of a corner is greater than the maximum value of self similarity 1%, be true, Mark it in red 
cv2.imshow('dst',img)
cv2.waitKey(0)
cv2.destoryAllWindows()