Calculates a feature map for corner detection

dst = cv.preCornerDetect(src)
dst = cv.preCornerDetect(src, 'OptionName', optionValue, ...)

Input

• src Source single-channel 8-bit of floating-point image.

Output

• dst Output image that has the single type and the same size as src.

Options

• KSize Aperture size of the cv.Sobel operator. default 3.
• BorderType Pixel extrapolation method. See cv.copyMakeBorder. default 'Default'

The function calculates the complex spatial derivative-based function of the source image.

dst = (Dx*src)^2*Dyy*src + (Dy*src)^2*Dxx*src - 2*Dx*src*Dy*src*Dxy*src

where Dx, Dy are the first image derivatives, Dxx, Dyy are the second image derivatives, and Dxy is the mixed derivative.

The corners can be found as local maximums of the functions, as shown below:

corners = cv.preCornerDetect(image, 'KSize',3);

% dilation with 3x3 rectangular structuring element
dilated_corners = cv.dilate(corners);
corner_mask = (corners == dilated_corners);