Abstract
We introduce a family of piecewise-exponential functions that have the Hermite interpolation property. Our design is motivated by the search for an effective scheme for the joint interpolation of points and associated tangents on a curve with the ability to perfectly reproduce ellipses. We prove that the proposed Hermite functions form a Riesz basis and that they reproduce prescribed exponential polynomials. We present a method based on Green's functions to unravel their multi-resolution and approximation-theoretic properties. Finally, we derive the corresponding vector and scalar subdivision schemes, which lend themselves to a fast implementation. The proposed vector scheme is interpolatory and level-dependent, but its asymptotic behaviour is the same as the classical cubic Hermite spline algorithm. The same convergence properties---i.e., fourth order of approximation---are hence ensured.
Introduction
In computer graphics and geometric modeling, Cubic Hermite splines are a standard tool for defining curves via "anchor points" and their associated "tangents" (slopes). While mathematically elegant and offering fourth-order approximation, they possess a significant limitation: they cannot perfectly represent circles or ellipses. Standard polynomial splines require a high number of control points to approximate "round" shapes, which complicates manual editing and automated tracking in fields like biomedical imaging.
This paper addresses this gap by introducing a new family of piecewise-exponential splines. These splines retain the attractive properties of Hermite splines—local control, smoothness, and fast computation—while adding the ability to represent ellipses exactly with a minimal number of points.
Methodology
The proposed framework departs from traditional polynomial spaces, utilizing a four-dimensional space of exponential polynomials: $\mathcal{E}_4 = \{1, x, e^{i\omega_0 x}, e^{-i\omega_0 x}\}$. This space naturally includes constants, linear trends, and trigonometric functions, which are the fundamental building blocks of circles and ellipses.
Hermite Interpolation and Basis Functions
The authors derive two specific basis functions, $\phi_1$ and $\phi_2$, that satisfy joint interpolation conditions at integer knots. These functions ensure that the resulting curve respects both the values and derivatives at the specified points.
Technical Components
- Green’s Functions: By using Green’s functions of specific differential operators, the authors link these new splines to standard exponential B-splines, proving their mathematical robustness.
- Subdivision Schemes: The researchers derive non-stationary "refinement masks." Because the underlying mathematics depends on the frequency parameter $\omega_0$, the subdivision rules change slightly at each level of refinement, allowing for efficient, iterative curve generation.
- Bézier Connection: The curves can be converted into an exponential version of Bernstein polynomials, enabling intuitive manipulation via Bézier-style handles.
Key Insights and Results
The proposed splines offer the "best of both worlds." They are asymptotically equivalent to classical cubic Hermite splines; as the grid is refined ($h \to 0$), the exponential functions converge to standard cubic polynomials. This ensures that the scheme maintains high-order accuracy ($O(h^4)$) while providing perfect ellipse reproduction at coarse levels.
Key findings include:
- Minimal Support: The basis functions have a local support of size 2, meaning local edits only affect a small portion of the curve.
- Stability: The functions form a Riesz basis, ensuring numerical stability where small changes in control points do not lead to unpredictable curve behavior.
- Efficiency: The derived vector subdivision scheme allows for fast rendering, making the approach suitable for real-time applications. The authors note that this model has been successfully applied to real-world biomedical image segmentation, such as outlining biological structures like cells or organs.
Limitations and Future Work
The primary trade-off is the non-stationarity of the subdivision scheme, which is slightly more complex to implement than stationary ones. Additionally, the frequency parameter $\omega_0$ must be tuned to the expected "roundness" of the target shape. While the computational overhead of trigonometric functions is negligible on modern hardware, future work could focus on automated parameter tuning for arbitrary shapes.
Summary updated