| DEVELOPMENT OF AN HYBRID CLASSIFICATION METHOD FOR HIGH SPATIAL RESOLUTION AERIAL IMAGES USING SHANNON ENTROPY |
| Paper ID : 1234-SMPR |
| Authors: |
|
mehdi mousavi *1, hamid ebadi2, abbas kiani2 1khaje nasir universiry, photogrammetry and remote sensing, tehran, iran 2Khajeh nasir university, tehran, iran |
| Abstract: |
| High spatial resolution remote sensing image classification is one of the most common ways to take advantage of the large volume of information contained in this type of images. Different methods have been used for this purpose in various researches. One of the most effective ways in this regard is to integrate two methods with different feature extraction perspective and complementary characteristics. The purpose of this paper is to provide an hybrid method based on convolutional neural network (CNN) and multilayer perceptron (MLP) for classifying aerial images. The decision fusion rules were designed based on the Shannon entropy of CNN predictions. The proposed ensemble classifier depicts the complementary outputs acquired from the CNN based on contextual representation and from MLP based on spectral features. The results of proposed method showed increase in all 3 evaluation criteria (precision, recall and f1-score) in comparison to individual CNN and MLP in test images. |
| Keywords: |
| Convolutional Neural Network, Multilayer Perceptron, Shannon Entropy, Classification, Deep Learning, Aerial Remote Sensing Imagery. |
| Status : Paper Accepted (Poster Presentation) |
