Segmentation of scanning-transmission electron microscopy images using the ordered median problem

This paper presents new models for segmentation of 2D and 3D Scanning-Transmission Electron Micro- scope images based on the ordered median function. The main advantage of using this function is its good adaptability to the different types of images to be studied due to the wide range of weight vec-...

Descripción completa

Detalles Bibliográficos
Autores: Calvino, José J., López Haro, Miguel, Muñoz Ocaña, Juan M., Puerto Albandoz, Justo, Rodríguez Chía, Antonio Manuel
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2022
País:España
Institución:Universidad de Sevilla (US)
Repositorio:idUS. Depósito de Investigación de la Universidad de Sevilla
OAI Identifier:oai:idus.us.es:11441/134912
Acceso en línea:https://hdl.handle.net/11441/134912
https://doi.org/10.1016/j.ejor.2022.01.022
Access Level:acceso abierto
Palabra clave:Location
Ordered median function
Segmentation
Clustering
Mixed integer linear programming
Descripción
Sumario:This paper presents new models for segmentation of 2D and 3D Scanning-Transmission Electron Micro- scope images based on the ordered median function. The main advantage of using this function is its good adaptability to the different types of images to be studied due to the wide range of weight vec- tors that can be cast. Classical segmentation models stand out for their ability to provide a segmentation of the original image very quickly and with low computational burden. However, they do not usually achieve high quality segmentations with a small number of clusters in order to classify the different ele- ments which compose the structure represented in the image. The quality of the segmentation provided by our approach is analysed using different choices of the weight vector in some real instances. More- over, improvements are proposed for the formulations to reduce the computational time needed to solve these problems by taking advantage of the weight vector structure.