FlowFilterSimple

A single pyramid level version of the filtering algorithm for computing optical flow.

A single pyramid level version of the filtering algorithm for computing optical flow.

This node can be used as a stand-alone algorithm when the optical flow magnitude is expected to be small (~2pix). Otherwise, it is recommended to use the FlowFilter node for a pyramidal version of the algorithm.

Parameters

gamma : float. Defaults to 0.01 The filter gains for the update step.

max_flow : float. Defaults to 1.0 The max magnitude allowed for the optical flow output.

smooth_iterations : int. Defaults to 1. The number of smooth iterations to apply to the estimated flow.

float_precision : int. Defaults to ll.FloatPrecision.FP32. Floating point precision used accross the algorithm. The outputs out_gray and out_flow will be of this floating point precision.

Inputs

in_gray : ImageView r8ui image. The input gray-scale image.

Outputs

out_gray : ImageView {r16f, r32f} image. The gray-scale image after one iteration of the algorithm. The image format depends on the float_precision parameter.

out_flow : ImageView {rg16f, rg32f} image. The estimated optical flow. The image format depends on the float_precision parameter.