Returns filter coefficients for computing spatial image derivatives

[kx, ky] = cv.getDerivKernels('OptionName', optionValue, ...)

Output

• kx Output matrix of row filter coefficients. It has the type KType.
• ky Output matrix of column filter coefficients. It has the type KType.

Options

• Dx Derivative order in respect of x. default 1
• Dy Derivative order in respect of y. default 1
• KSize Aperture size. It can be 'Scharr', 1, 3, 5, or 7. default 3.
• Normalize Flag indicating whether to normalize (scale down) the filter coefficients or not. Theoretically, the coefficients should have the denominator = 2^(KSize*2-Dx-Dy-2). If you are going to fitler floating-point images, you are likely to use the normalized kernels. But if you compute derivatives of an 8-bit image, store the results in a 16-bit image, and wish to preserve all the fractional bits, you may want to set Normalize=false. default false
• KType Type of filter coefficients. It can be single or double. default single

The function computes and returns the filter coefficients for spatial image derivatives. When KSize='Scharr', the Scharr 3x3 kernels are generated (see cv.Scharr). Otherwise, Sobel kernels are generated (see cv.Sobel). The filters are normally passed to cv.sepFilter2D.