This title appears in the Scientific Report :
2018
Please use the identifier:
http://dx.doi.org/10.1109/IGARSS.2018.8518671 in citations.
Please use the identifier: http://hdl.handle.net/2128/19910 in citations.
The Influence of Sampling Methods on Pixel-Wise Hyperspectral Image Classification with 3D Convolutional Neural Networks
The Influence of Sampling Methods on Pixel-Wise Hyperspectral Image Classification with 3D Convolutional Neural Networks
Supervised image classification is one of the essential techniques for generating semantic maps from remotely sensed images.The lack of labeled ground truth datasets, due to the inherent time effort and cost involved in collecting training samples, has led to the practice of training and validating...
Saved in:
Personal Name(s): | Lange, Julius |
---|---|
Cavallaro, Gabriele / Götz, Markus / Ernir, Erlingsson / Riedel, Morris | |
Contributing Institute: |
Jülich Supercomputing Center; JSC |
Published in: |
IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium |
Imprint: |
2018
|
Physical Description: |
2087 - 2090 |
DOI: |
10.1109/IGARSS.2018.8518671 |
Conference: | IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Valencia (Spain), 2018-07-22 - 2018-07-27 |
Document Type: |
Contribution to a book Contribution to a conference proceedings |
Research Program: |
Upgrade of CaSToRC into a Center of Excellence in Simulation and Data Science DEEP - Extreme Scale Technologies Data-Intensive Science and Federated Computing |
Link: |
OpenAccess OpenAccess |
Publikationsportal JuSER |
Please use the identifier: http://hdl.handle.net/2128/19910 in citations.
Supervised image classification is one of the essential techniques for generating semantic maps from remotely sensed images.The lack of labeled ground truth datasets, due to the inherent time effort and cost involved in collecting training samples, has led to the practice of training and validating new classifiers within a single image. In line with that, the dominant approach for the division of the available ground truth into disjoint training and test sets is random sampling. This paper discusses the problems that arise when this strategy is adopted in conjunction with spectral-spatial and pixel-wise classifiers such as 3D Convolutional Neural Networks (3D CNN). It is shown that a random sampling scheme leads to a violation of the independence assumption and to the illusion that global knowledge is extracted from the training set.To tackle this issue, two improved sampling strategies based on the Density-Based Clustering Algorithm (DBSCAN) are proposed. They minimize the violation of the train and test samples independence assumption and thus ensure an honest estimation of the generalization capabilities of the classifier. |