ryokamoi / implicit_curve_approximation

Implementation of implicit curve approximation methods

Imlcit Curve Approximation

Environments

• Python 3.6.2
• libraries are listed in pip.freeze

How To Run

python main.py configfile.txt

Config File

Example for gradient-one method. Please refer to "config" directory for other methods and more details.

model=tasdizen2000
dataset=dataset/shape1
degree=4
mu=1.0

Methods

Classic Least Square

• G. Taubin, 1991 (Estimation of Planar Curves, Surfaces, and Nonplanar Space Curves Defined by Implicit Equations with Applications to Edge and Range Image Segmentation)
• "model=taubin1991"

python main.py config/taubin1991_shape1.txt

Gradient-One Method

• T. Tasdizen, et al., 2000 (Improving the Stability of Algebraic Curves for Applications)
• "model=tasdizen2000"
• The data points should be ordered in this implementation (there are alternative way to calculate local gradient if the data points are not ordered)

python main.py config/tasdizen2000_shape1.txt

Topologically Faithful Fitting

• D. Keren, et al., 2004 (Topologically Faithful Fitting of Simple Closed Curves)
• "model=keren2004"
• In this code, Naive Nelder-Mead method (from scipy) is used
• The estimation for the following result does not use 3L algorithm. 3L algorithm is implemented in our code, but it does not improve fitting very much in my implementation.

python main.py config/keren2004_shape1.txt

Author

Ryo Kamoi (https://ryokamoi.github.io/)