Image registration based on kernel-predictability

In this work, a new similarity measure between images is presented, which is based on the concept of predictability of random variables evaluated through kernel functions. Image registration is achieved maximizing this measure, analogously to registration methods based on entropy, like mutual inform...

Descripción completa

Detalles Bibliográficos
Autor: JOSE LUIS MARROQUIN ZALETA
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2008
País:México
Institución:Centro de Investigación en Matemáticas
Repositorio:Repositorio Institucional CIMAT
Idioma:inglés
OAI Identifier:oai:cimat.repositorioinstitucional.mx:1008/946
Acceso en línea:http://cimat.repositorioinstitucional.mx/jspui/handle/1008/946
Access Level:acceso abierto
Palabra clave:info:eu-repo/classification/MSC/Kernel - Lenguaje de Programación
info:eu-repo/classification/cti/1
info:eu-repo/classification/cti/12
info:eu-repo/classification/cti/1203
info:eu-repo/classification/cti/120302
Descripción
Sumario:In this work, a new similarity measure between images is presented, which is based on the concept of predictability of random variables evaluated through kernel functions. Image registration is achieved maximizing this measure, analogously to registration methods based on entropy, like mutual information and normalized mutual information. Compared experimentally with these methods in different problems, our proposal exhibits a more robust performance specially for problems involving large transformations and in cases where the registration is done using a small number of samples, such as in nonparametric registration.