Spatially continuous dual optimization on compactness function for image segmentation

doi:10.1016/j.patcog.2025.112613

Spatially continuous dual optimization on compactness function for image segmentation

2025-10-29

0

PRE

AI

DOI：10.1016/j.patcog.2025.112613

Original

Original request for help

Save

Abstract

En 中文

• A novel spatially continuous dual optimization model for image segmentation that incorporates a compactness function for preserving shape priors. • Development of a soft threshold dynamics (STD)-based primal-dual algorithm that offers improved numerical stability and computational efficiency. • The approach demonstrates insensitivity to initialization parameters, functioning effectively as both post-processing and standalone segmentation. • Integration with convex shape priors creates an iterative framework that simultaneously achieves smooth boundaries and refined object shapes. • Extensive experiments validate the approach’s versatility and superior performance on both medical imaging and natural scene datasets for various segmentation tasks

Spatially continuous dual optimization on compactness function for image segmentation

Abstract

Journal

Organization