calcGradients

Calculate gradient magnitude and orientation of an image slice.

void

calcGradients

(

InputTensor

V = DeepElementType!InputTensor

)

(

InputTensor input

ref Slice!(Contiguous, [2], V*) mag

ref Slice!(Contiguous, [2], V*) orient

EdgeKernel edgeKernelType = EdgeKernel.SIMPLE

TaskPool pool = taskPool

)

if (

isFloatingPoint!V

)

in { static assert (isSlice!InputTensor, "Input tensor has to be of type mir.ndslice.slice.Slice"); static assert (InputTensor.init.shape.length == 2, "Input tensor has to be 2 dimensional. (matrix)"); assert (!input.empty); }

Parameters

input: Type: InputTensor
Input slice of an image.
mag: Type: Slice!(Contiguous, [2], V*)
Output magnitude value of gradients. If shape does not correspond to input, is allocated anew.
orient: Type: Slice!(Contiguous, [2], V*)
Orientation value of gradients in radians. If shape does not correspond to input, is allocated anew.
edgeKernelType: Type: EdgeKernel
Optional convolution kernel type to calculate partial derivatives. Default value is EdgeKernel.SIMPLE, which calls calcPartialDerivatives function to calculate derivatives. Other options will perform convolution with requested kernel type.
pool: Type: TaskPool
TaskPool instance used parallelise the algorithm.
Note: Input slice's memory has to be contiguous. Magnitude and orientation slices' strides have to be the identical.

calcGradients

Parameters

Meta

Source