1

I know that the gradient (Gx,Gy) is to measure the largest change direction of image function G(x,y). Therefore, the edge detection is to find maximum points of Gx^2+Gy^2. Thus, what's the principle of Laplacian edge detection? Thanks for any help.

hippietrail
  • 15,848
  • 18
  • 99
  • 158
Hooben
  • 451
  • 1
  • 4
  • 6

1 Answers1

0

To be more precise, it is the zero-crossings of the Laplacian, which are used to detect edges. The Laplacian operator is a function of the second derivatives of the image. An edge is a local maximum of the first derivative, which is the point where the second derivative is zero.

Dima
  • 38,860
  • 14
  • 75
  • 115
  • In the case of dimension one, a local maximum of the first derivate is just the zeros of the second derivative. However, in dimesion 2, it's not obvious. I still do not understand why the crossing zeros of the Laplacian denote edge. – Hooben Aug 31 '16 at 15:39