文章基本信息

标题：Terrain Classification Using Static and Dynamic Texture Features by UAV Downwash Effect
本地全文：下载
作者：João Pedro Carvalho ; José Manuel Fonseca ; André Damas Mora 等
期刊名称：Journal of Automation, Mobile Robotics & Intelligent Systems (JAMRIS)
印刷版ISSN：1897-8649
电子版ISSN：2080-2145
出版年度：2019
卷号：13
期号：1
页码：84-93
DOI：10.14313/JAMRIS_1-2019/11
出版社：Industrial Research Inst. for Automation and Measurements, Warsaw
摘要：􀀼nowing how to iden􀆟fy terrain types is especially importantin the autonomous naviga􀆟on, mapping, decisionmaking and emergency landings areas. For example, anunmanned aerial vehicle (UAV) can use it to find a suitablelanding posi􀆟on or to cooperate with other robotsto navigate across an unknown region. Previous workson terrain classifica􀆟on from RGB images taken onboardof UAVs shown that only sta􀆟c pixel-based features weretested with a considerable classifica􀆟on error. This paperpresents a computer vision algorithm capable of iden􀆟-fying the terrain from RGB images with improved accuracy.The algorithm complement the sta􀆟c image featuresand dynamic texture pa􀆩erns produced by UAVs rotorsdownwash e􀄫ect (visible at lower al􀆟tudes) and machinelearning methods to classify the underlying terrain.The system is validated using videos acquired onboard ofa UAV with a RGB camera.
关键词：Image processing; Texture; Machine Learning;Terrain 􀀒lassifica􀆟on; 􀁅eural 􀁅etworks; UAV