- Areas in image where image brightness changes sharply
- Uses methods that approximates derivatives
- When taking difference between 2 consecutive pixels,
the positive value represents the directional vector of edge
- When taking difference between 2 consecutive pixels,
the negative value represents the opposite directional vector of edge
- We can also get the same result bu convolving the pictures with
special kernels or filters (sobel operators)
- Combining the results of all convolution will give us
an approximation of gradient vector in all direction
- Magnitude of the gradient gives us the intensity value, which is
an image that represent edges
- Angle of the gradient gives us the direction of edges