Rodolfo Lotte

Company Owner @ LCIARS / Consultant in Remote Sensing, Machine Learning and Computer Vision / PhD in Remote Sensing / MSc. in Applied Computing

Roads center-axis extraction in airborne SAR images: an approach based in active contour model with the use of semi-automatic seeding

Download dissertation Download defense Related paper

Abstract

The research involving computational methods for the extraction of roads has intensified in the last two decades. The procedure is usually performed through the analysis of images acquired by optical sensors or radar (Radio Detection and Ranging). The great advantage in the usage of radar images is the possibility of survering areas often covered by clouds, since the imaging by active sensors is independent from atmospheric conditions in the region of interest. The cartographic mapping based on these images is often done manually, requiring considerable time and effort from the interpreter. There are currently many studies involving the extraction of roads by means of automatic or semi-automatic approaches, however, each of them has different solutions for different problems, making this task a scientific issue still open. Among the advantages of using radar images, one can mention the acquistion of images regardless of atmospheric and ilumination conditions, besides the possibility of surveying regions where the terrain is hidden by the vegetation canopy, among others. This work comprises the semi-automatic generation of seed points located dose to the features of interest by means of Self-Organizing Maps (SOM). We then employ an active contour method (Snakes) for the extraction of the roads centreaxis in a synthetic aperture radar (SAR) airborne image. The obtained results were evaluated as to their quality with respect to perfection, correction, and redundancy.

Info! Dear researcher/scientist/academic, you may find some Wikipedia references in this page, which are totally directed for those that might not be familiar with terms and need a more illustrative and didatical understanding. Please, fell free to contribute at any moment.

Contextualization

Source-code and tools

Coding:


MATLAB	C++	Java

IDE:


ENVI	Netbeans

Arts:


Illustrator	Photoshop

Documentation

	$alt-LaTeX$
Kile	LaTeX

Presentation

image-left The board of examiners was composed by Prof. Dr. Thales Körting (INPE), Prof. Dr. Norbert Haala (IfP-Sttutgart), Prof. Dr. Edson Mitishita (UFPR), Prof. Dr. Antônio Maria G. Tommaselli (UNESP-Prudente), Dr. Fabien H. Wagner (INPE), Dr. Luiz E. O. E. Aragão (INPE-Advisor), and Dr. Yosio E. Shimabukuro (INPE-Advisor).

The presentation was unanimously approved and is available physically for consultation in the collection of INPE, in São José dos Campos, in the digital library, or simply by clicking the orange button above.

Cite this dissertation

@MastersThesis{Lotte2012,
               author = "Lotte, Rodolfo Georjute",
                title = "Roads center-axis extraction in airborne SAR images: an approach based 
         in active contour model with the use of semi-automatic seeding",
               school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
                 year = "2012",
              address = "São José dos Campos",
                month = "2012-12-07",
             keywords = "extração de estradas, imagem de radar, modelo de contorno ativo, semeação semiautomática,
                          extraction of roads, radar image, activr contour method,
                          semiautomatic seeding.",
            committee = "Dutra, Luciano Vieira (presidente) and Sant'Anna, Sidnei João Siqueira (orientador) 
            and Almeida, Cláudia de (orientadora) and Costa, Gilson Alexandre Ostwald Pedro",
           copyholder = "SID/SCD",         
             language = "pt",
                pages = "160",
}