59893

Comparativo entre os algoritmos K-Means e ckMeans para mapeamento automatizado de uso do solo

Favoritar este trabalho

Remote sensing allow us to acquire information about an object or phenomenon without the need to make physical contact with the object, which turn it usable in many fields, like hydrology, ecology, oceanography, glaciology, geology. Remote Sensing generally refers to the use of satellite-based (or aircraft) sensor technologies to detect and classify objects on Earth. Classification is the process of extracting information in images (or data) to recognize patterns and homogeneous objects and are used in remote sensing to map areas of the earth''s surface. This article makes a comparison between two algorithms used in image classification applied to remote sensing. The first one is the well-known K-Means, that has the characteristic to be fast and its modeling is relatively simple, and the second is the fuzzy ckMeans algorithm that allows to model inaccurate data according to their membership degree. The ckMeans algorithm proved to be a good alternative in the image segmentation process. To validate the work we compared the classification of an image, obtained by a satellite, of the western border of the state of Rio Grande do Sul and defined a priori four clusters. Then, the classification between K-Means and ckMeans algorithms was performed. Finally, a domain knowledge specialist discussed the resultant classification obtained by these algorithms.