Spline

C++/CUDA library for manipulating 2D parametric B-splines.

Features

• Creation splines from a vector of points
• Creation of spiral-like splines
• Creation of ellipses
• Finding intersection points of two curves (splines or ellipses) with an accuracy of 1e-9 using Newton-Raphson optimization algorithm
• Finding the closest points of two curves (splines or ellipses) with an accuracy 1e-9 using RMSProp optimization or BruteForce search algorithms
• C++/CUDA and C++/OMP algorithms implementation

Speed estimation

A high computation load example of a spline and an ellipse closest points finding:

RMSProp CPU	~ 3300 us **
BruteForce CPU	~ 2600 us **
BruteForce CUDA	~ 790 us ***

** with Intel i5-13600KF

*** with NVIDIA GeForce RTX 3060 Ti

Building

Linux:

$ git clone https://github.com/Igkho/Spline.git
$ cd Spline
$ mkdir build
$ cd build
$ cmake ..
$ make -j10

Windows:

• run CMake-gui;
• specify the source code and binary directories;
• press Configure button, set up the Generator settings (i.e. select the MSVC version);
• press Generate button;
• press Open project button to open solution in MSVC and build the solution;

Executables:

• Spline - the spline manipulation library
• SplineTests - unit tests of the library
• SplineExample - generates splines intersection and closest points finding examples

Documentation:

doxygen html documentation is generated in docs/html

Dependencies:

• the library depends on Eigen (fetched by Cmake), Nvidia CUDA, OpenMP (optional)
• tests depend on GoogleTest (fetched by Cmake)
• examples depend on OpenCV (to run the executable the OpenCV libraries should be on the search path)
• documentation building depends on Doxygen and GraphViz

Examples

Intersecting two splines:

Intersecting a spline and an ellipse:

Intersecting spirals:

Closest points of two splines:

Closest points of a spline and an ellipse: