DeepFlow optical flow algorithm implementation

flow = cv.calcOpticalFlowDF(I0, I1)

Input

• I0 First 8-bit single-channel grayscale input image.
• I1 Second input image of the same size and type as I0.

Output

• flow computed flow image that has the same size as I0 and type single (2-channels). Flow for (x,y) is stored in the third dimension.

The class implements the DeepFlow optical flow algorithm described in [Weinzaepfel2013]. See also DeepMatching. Parameters - class fields - are:

• Alpha Smoothness assumption weight. default 1.0
• Delta Color constancy assumption weight. default 0.5
• Gamma Gradient constancy weight. default 5.0
• Sigma Gaussian smoothing parameter. default 0.6
• MinSize Minimal dimension of an image in the pyramid (next, smaller images in the pyramid are generated until one of the dimensions reaches this size). default 25
• DownscaleFactor Scaling factor in the image pyramid (must be <1). default 0.95
• FixedPointIterations How many iterations on each level of the pyramid. default 5
• SorIterations Iterations of Successive Over-Relaxation (solver). default 25
• Omega Relaxation factor in SOR. default 1.6

References

[Weinzaepfel2013]:

Philippe Weinzaepfel, Jerome Revaud, Zaid Harchaoui, and Cordelia Schmid. "Deepflow: Large displacement optical flow with deep matching". In Computer Vision (ICCV), 2013 IEEE International Conference on, pages 1385-1392. IEEE, 2013.