Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces a multiscale kinetic framework for image segmentation by modeling images as interacting particle systems characterized by spatial position and color features. They derive a kinetic formulation for particle density combining transport, aggregation, and diffusion, and then obtain a macroscopic model describing the evolution of pixel fractions with certain features. The framework uses particle-based optimization techniques based on this reduced-complexity model, demonstrating effectiveness and robustness to noise in numerical tests.
Bridging particle systems and continuum models offers a new, robust approach to image segmentation that handles noisy data effectively.
In this work, we present a multiscale kinetic framework for consensus-based image segmentation. By interpreting an image as a system of interacting particles, each pixel is characterised by its spatial position and an internal feature encoding color information. We introduce a coupled interaction scheme governing the evolution of particles in both position and feature spaces, from which we derive a kinetic formulation for the particle density in the space-feature domain combining transport, aggregation, and diffusion effects. Furthermore, through a suitable scaling, we obtain a first-order macroscopic model describing the evolution of the fraction of pixels carrying information on the fraction of pixels having a certain feature. Based on this reduced-complexity model, we present a data-oriented approach where we make use of particle-based optimisation techniques for the accurate segmentation of images. Numerical tests show the effectiveness of the proposed framework and its robustness under different noise conditions.