CLASSIFICATION AND DETECTION OF HYPERSPECTRAL IMAGES

Abstract

Recently, many neural network has been extensively applied to hyperspectral image (HSI) for
classification and they had shown promising results on various visual tasks. However, its success is greatly
attributed to numerous labeled samples, whose acquisition costs a large amount of time and money. In order
to improve the classification performance while reducing the labeling cost, we are using convolutional neural
network for classification of hyperspectral Images. Both spectral and spatial features are considered for
classification of hyperspectral images. Mostly, spectral features are used for classification to improve the
accuracy of the classification. Convolutional neural network achieves the highest accuracy of 98% in the
classification of hyperspectral images compared with other methods. In this, classification process we are
using principle component. PCA is a very powerful technique for hyperspectral classification. PCA cuts down
the calculation time of classification by the significant amount and also reduces the amount of data to be
handled. The PCA preprocessing gives rather acceptable and accurate classification results. The main
purpose of classification of hyperspectral images is to assign a class label to each pixel. And calculation of
accuracy of classification and displaying the classified image. The results of the classification model, we get
many images such as ground truth image, preprocessing image, gray scale image, training map, testing
image, classified image and segmented image. The classified is obtained at the 74% accuracy and the
segmented image is obtained at 83% accuracy.