:: Volume 5, Issue 3 (12-2017) ::
jgit 2017, 5(3): 77-97 Back to browse issues page
Comprehensive investigation on non-parametric classification methods in order to separate urban objects using the integration of very high spatial resolution LiDAR and aerial data
Zinat Ghavami, Hossein Arefi *, Behnaz Bigdeli, Milad Janalipour
University of Tehran
Abstract:   (4327 Views)
Nowadays, to obtain information covering urban land, the city is one of the most important and widely used management tools in the study of Earth changes. Classification of images is one of the most common methods of extracting information from remote sensing data. Complex and dense urban areas are one of the problems in the analysis of remote sensing. The accuracy of classification performance in these areas is under the attention of the researchers and always tries to improve the accuracy. Using different data and application integration techniques to classify a variety of effects can be more accurate-achieved with higher reliability. Among the successful classification methods in recent years, support vector machines algorithm and ensemble learning algorithms such as Bagging, Boosting and Random Forest noted can be mentioned. In this paper, the performance of the four algorithms to identify the effects of the dense city with a very high resolution aerial LIDAR and image are discussed. The results show that the combination of LIDAR data and aerial image, gives out a better classification of the degree of urban features. The classification of urban features with the help of integrated LIDAR and aerial image information with the use of support vector machine algorithm-precision 93.99% performs higher ability than other classification methods such as Bagging, Boosting and Random Forest.
Keywords: LIDAR Data, Areal Image, Support Vector Machines, Bagging, Boosting, Random Forest
Full-Text [PDF 2378 kb]   (2179 Downloads)    
Type of Study: Research | Subject: RS
Received: 2016/07/4 | Accepted: 2017/05/30 | Published: 2018/01/10

XML   Persian Abstract   Print

Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Volume 5, Issue 3 (12-2017) Back to browse issues page