Line Integral Convolution

The Line Integral Convolution (LIC) is an algorithm used to image a vector field. Its main advantage is to show in intricate detail the fine structure of the vector field. It does not display the direction or magnitude of the vectors, although this information can be color coded in a postprocessing step.

The result of course depends on the shape of the kernel and the length
of the streamline. If it is too small, the texture is not sufficiently filtered and the motion is not clear. If it is too large, the image is smoothed and details of the motion are lost. For an image of size (256, 256), a value of 20 provides acceptable results.

Install

If you want to install LIC you can clone the repo and run.

    pip install -e .

or install from pypi

    pip install licplot

Usage

The basic usage is shown in and a runnable example can be found under examples/lic_demo.py

    from lic import lic_internal
    import numpy as np
    import matplotlib.pyplot as plt
    # create vector field and kernel
    size = 500
    u = np.zeros((size, size), dtype=np.float32)
    v = np.zeros((size, size), dtype=np.float32)
    texture = np.random.rand(size, size).astype(np.float32)

    # create a kernel
    kernel_length = 31
    kernel = np.sin(np.arange(kernel_length) * np.pi / kernel_length).astype(np.float32)

    # compute the lic
    image = lic_internal.line_integral_convolution(u, v, texture, kernel)

    plt.imshow(image, cmap="hot")
    plt.show()

Forked from https://github.com/aarchiba/scikits-vectorplot

by Anne Archibald