Artificial Neural Network Approach for Land Cover Classification of Fused Hyperspectral and Lidar Data

Abstract : Hyperspectral remote sensing images are consisted of several hundreds of contiguous spectral bands that can provide very rich information and has the potential to differentiate land cover classes with similar spectral characteristics. LIDAR data gives detailed height information and thus can be used complementary with Hyperspectral data. In this work, a hyperspectral image is combined with LIDAR data and used for land cover classification. A Principal Component Analysis (PCA) is applied on the Hyperspectral image to perform feature extraction and dimension reduction. The first 4 PCA components along with the LIDAR image were used as inputs to a supervised feedforward neural network. The neural network was trained in a small part of the dataset (less than 0.4%) and a validation set, using the Bayesian regularization backpropagation algorithm. The experimental results demonstrate efficiency of the method for hyperspectral and LIDAR land cover classification.
Document type :
Conference papers
Complete list of metadatas

Cited literature [9 references]  Display  Hide  Download

https://hal.inria.fr/hal-01459617
Contributor : Hal Ifip <>
Submitted on : Tuesday, February 7, 2017 - 1:05:10 PM
Last modification on : Friday, December 1, 2017 - 1:16:34 AM
Long-term archiving on : Monday, May 8, 2017 - 2:11:05 PM

File

978-3-642-41142-7_26_Chapter.p...
Files produced by the author(s)

Licence


Distributed under a Creative Commons Attribution 4.0 International License

Identifiers

Citation

Paris Giampouras, Eleni Charou, Anastasios Kesidis. Artificial Neural Network Approach for Land Cover Classification of Fused Hyperspectral and Lidar Data. 9th Artificial Intelligence Applications and Innovations (AIAI), Sep 2013, Paphos, Greece. pp.255-261, ⟨10.1007/978-3-642-41142-7_26⟩. ⟨hal-01459617⟩

Share

Metrics

Record views

467

Files downloads

301