Revista Digital Universitaria ISSN: 1607 - 6079 | Publicación mensual | 1 de junio de 2015 vol.16, No.6
ABSTRACT
Satellite and radar image processing is the starting point in the geomorphological mapping and aims to characterize objects by shape, structure or color. This paper presents these images correcting as the first step in the detection of impact craters, such as BP and Oasis (Libya), which have been characterized by analyzing radar images (Radarsat-1) and used as standards of reference to differentiate impact craters from those of volcanic origin. The analysis of Radarsat-1 imagery consists of: filtering (known as speckle), segmenting (partition of the image into regions with common characteristics) and classifying (grouping of pixels based on selected features). About this latter action, this paper proposes automatic processes that classify, from imagery data corrected from filtering and segmentation, the geomorphic objects into two groups: impact and volcanics. The relevance of this classifier computer system is that it acts with Machine Learning based on artificial neural networks (ANN).
Keywords: Radarsat-1, impact craters, BP, Oasis, speckle effect, segmentation, ANN classification, machine learning
Digital Image Processing Applied to Morphological Recognition of Impact Structures in the Earth's Surface
Nancy Jiménez, Raquel Díaz, Silvia Ambrocio, Màrius RamírezKeywords: Radarsat-1, impact craters, BP, Oasis, speckle effect, segmentation, ANN classification, machine learning