Abstract
Land cover classification analysis from satellite imagery methods are important because they are the basis for characterizing surface conditions and evolution, supporting the management and optimization of land resources, evaluating global climate and environmental changes, and facilitating sustainable regional economic and social development. In order to address these necessities, artificial neural networks have been used extensively. In addition, other methods based on computer vision are very useful to solve this task. In this paper, the authors propose an approach based on Monte Carlo method and artificial neural networks in order to classify regions of small forest reserves from drones’ images and calculate their respective areas. Next to the small forest reserve will be extended a standard rectangular tarpaulin of 250 square meters and based on this reference it will be possible to calculate the area of the forest reserve if the ground is relatively flat. The proposed approach will be compared with a method based on watershed algorithm. The automatic calculation of the forest area through images generated by drones has much practical application for environmental engineers, for example, for the calculation of environmental impact and determination of carbon loss if such forests are consequently deforested.
Highlights
Land cover can be defined as a combination of natural and artificial surface structures occupying a given territory [1]
It will be compared an approach based on multilayer perceptron (MLP) and Monte Carlo methods with another one based on the watershed algorithm, when used to calculate forest reserve areas from photos captured by drones
We will start with a brief explanation about Artificial Neural Networks; we will summarize some concepts about Monte Carlo methods, and will finalize the section describing variations of Watershed algorithm
Summary
Land cover can be defined as a combination of natural and artificial surface structures occupying a given territory [1]. Methods to rapidly and accurately perform the extraction of land cover information using satellite remote sensing are important part of this research area [3, 4]. Such methods are important because they are the basis for characterizing surface conditions and evolution, base for supporting management and optimization of land resources, assessing global climate and environmental changes and facilitating. An approach using the watershed algorithm with markers and filters was used to delineate tree contours in [14] In this work, it will be compared an approach based on multilayer perceptron (MLP) and Monte Carlo methods with another one based on the watershed algorithm, when used to calculate forest reserve areas from photos captured by drones.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
