Abstract:
This paper proposes a network classification framework based on multiresolution local binary pattern techniques and convolutional neuronal network models. Local binary pa...Show MoreMetadata
Abstract:
This paper proposes a network classification framework based on multiresolution local binary pattern techniques and convolutional neuronal network models. Local binary pattern is used as a descriptor in order to evaluate the data in the spatial domain of the hyperspectral data. The convolutional neural networks offer powerful feature extraction capabilities over the spectral information defining the high-resolution data sets. Our proposed model is based on one dimensional convolutional network (1D-CNN) applied over the histograms obtained from the rotational invariant local binary pattern technique. With respect to other concurrent methods applied on textured hyperspectral data bases, the novel proposed method has similar results with other LBP-ID-CNN classification networks, but has smaller number of parameters and reduced time for training and testing. The experimental results demonstrate that the proposed method can improve the accuracy on Indian Pines and Salinas data sets even with a limited number of training samples.
Published in: 2020 IEEE International Conference on Automation, Quality and Testing, Robotics (AQTR)
Date of Conference: 21-23 May 2020
Date Added to IEEE Xplore: 01 July 2020
ISBN Information: