Publicação
Urban areas identification through clustering trials and the use of neural networks
| Resumo: | The main objective of this paper is to assess how an urban area can be identified accurately using satellite images. The case study is the urban centre of Vila Real and the satellite image used is SPOT5. Results are still being worked out and some shortcomings of the process are known. A most important one is the lack of large data sets as only one satellite image is normally used. This problem is typical in urban studies due to the funding shortage. Important decisions are concerned with the number of classes and their homogeneity as well as the estimation of accuracy. Following the obtained results, it will be possible to choose the most efficient classification in terms of performance, accuracy and confidence level upon working method. Likewise, the monitoring process of urban areas expansion can be used more extensively. |
|---|---|
| Autores principais: | Lourenço, Júlia |
| Outros Autores: | Ramos, L.; Ramos, Rui A. R.; Santos, Henrique Dinis dos; Fernandes, Delfim |
| Assunto: | Remote sensing Neural networks Maximum likelihood classification Supervised classification |
| Ano: | 2005 |
| País: | Portugal |
| Tipo de documento: | comunicação em conferência |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade do Minho |
| Idioma: | inglês |
| Origem: | RepositóriUM - Universidade do Minho |
| Resumo: | The main objective of this paper is to assess how an urban area can be identified accurately using satellite images. The case study is the urban centre of Vila Real and the satellite image used is SPOT5. Results are still being worked out and some shortcomings of the process are known. A most important one is the lack of large data sets as only one satellite image is normally used. This problem is typical in urban studies due to the funding shortage. Important decisions are concerned with the number of classes and their homogeneity as well as the estimation of accuracy. Following the obtained results, it will be possible to choose the most efficient classification in terms of performance, accuracy and confidence level upon working method. Likewise, the monitoring process of urban areas expansion can be used more extensively. |
|---|