Non-parametric edge detection in speckled imagery
AUTOR(ES)
Edwin Giovanny GirÃn Amaya
DATA DE PUBLICAÇÃO
2008
RESUMO
This thesis proposes a non-parametric technique for boundary detection in speckled imagery. Synthetic Aperture Radar (SAR), sonar, B-ultrasound and laser imagery is corrupted by a signal-dependent non-additive noise called speckle. Several statistical models have been proposed to describe such a noise, thus of specialized techniques for image improvement and analysis. The G0 distribution is a statistical model that succeeds in describing a wide range of areas as, for instance in SAR data, pastures (smooth), forests (rough) and urban (extremely rough) areas. The aim of this thesis is to develop alternative techniques for edge detection in speckled imagery. Its starting point are the works by Gambini et al. (2006, 2008). We describe a new edge detector based on the Kruskal Wallis test and show that it is an useful alternative to the method proposed by M. Gambini, which is based on the likelihood function of the data. We provide evidence that the M. Gambini technique can be successfully replaced by the Kruskal Wallis method. The latter is more computationally efficient, the corresponding algorithm being up to 1000 times fasted that the M. Gambini algorithm
ASSUNTO(S)
detecÃÃo de bordas imagens sar non-parametrics statistics edges detection imagens speckled sar imagery estatÃstica nÃo-paramÃtrica estatistica speckle noise speckled imagery ruÃdo speckle
Documentos Relacionados
- Non-parametric stability analyses of dough properties in wheat
- A non-parametric model for transcription factor binding sites
- On Edwards' criterion of seasonality and a non-parametric alternative.
- New insights into genomic selection through population-based non-parametric prediction methods
- On the measurement of capacity utilisation and coast efficiency: a non-parametric approach at firm level