TV-L1 vs iLK

Comparison the TV-L1 and iLK methods for the optical flow estimation.

Out:

Processing the Beanbags sequence
        TV-L1 processing time: 2.598048sec
        ILK processing time: 0.454992sec
Processing the Dimetrodon sequence
        TV-L1 processing time: 1.877484sec
        ILK processing time: 0.313538sec
Processing the Dogdance sequence
        TV-L1 processing time: 2.664790sec
        ILK processing time: 0.458793sec
Processing the Grove2 sequence
        TV-L1 processing time: 2.905357sec
        ILK processing time: 0.456460sec
Processing the Grove3 sequence
        TV-L1 processing time: 2.968026sec
        ILK processing time: 0.458371sec
Processing the Hydrangea sequence
        TV-L1 processing time: 2.124195sec
        ILK processing time: 0.313094sec
Processing the Minicooper sequence
        TV-L1 processing time: 2.884015sec
        ILK processing time: 0.457555sec
Processing the Rubberwhale sequence
        TV-L1 processing time: 2.235224sec
        ILK processing time: 0.312935sec
Processing the Urban2 sequence
        TV-L1 processing time: 2.663834sec
        ILK processing time: 0.457162sec
Processing the Urban3 sequence
        TV-L1 processing time: 2.621930sec
        ILK processing time: 0.458131sec
Processing the Venus sequence
        TV-L1 processing time: 1.429141sec
        ILK processing time: 0.228746sec
Processing the Walking sequence
        TV-L1 processing time: 2.456856sec
        ILK processing time: 0.458667sec

from time import time
import matplotlib.pyplot as plt
import pyimof


seq_list = pyimof.data.__all__
seq_count = len(seq_list)

fig = plt.figure(figsize=((9, 2*seq_count)))
ax_array = fig.subplots(seq_count, 3)
ax_array[0, 0].set_title("Reference image")
ax_array[0, 1].set_title("TV-L1 result")
ax_array[0, 2].set_title("iLK result")

# --- Loop over available sequences

for name, (ax0, ax1, ax2) in zip(seq_list, ax_array):

    title = name.capitalize()

    print(f"Processing the {title} sequence")

    # --- Load data
    I0, I1 = pyimof.data.__dict__[name]()

    ax0.imshow(I0, cmap='gray')
    ax0.set_ylabel(title)
    ax0.set_xticks([])
    ax0.set_yticks([])

    # --- Run TV-L1

    t0 = time()
    u, v = pyimof.solvers.tvl1(I0, I1)
    t1 = time()

    pyimof.display.plot(u, v, ax=ax1, colorwheel=False)

    print("\tTV-L1 processing time: {:02f}sec".format(t1-t0))

    # --- Run iLK

    t0 = time()
    u, v = pyimof.solvers.ilk(I0, I1)
    t1 = time()

    pyimof.display.plot(u, v, ax=ax2, colorwheel=False)

    print("\tILK processing time: {:02f}sec".format(t1-t0))

fig.tight_layout()

plt.show()