AUTHOR=Ye Fanghong , Zhou Zheng , Wu Yue , Enkhtur Bayarmaa 

TITLE=Application of convolutional neural network in fusion and classification of multi-source remote sensing data

JOURNAL=Frontiers in Neurorobotics

VOLUME=Volume 16 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/fnbot.2022.1095717

DOI=10.3389/fnbot.2022.1095717

ISSN=1662-5218

ABSTRACT=Through remote sensing images, we can understand and observe the terrain, and its application scope is relatively large, such as agriculture, military, etc. In order to achieve more accurate and efficient multi-source remote sensing data fusion and classification, DB-CNN algorithm is proposed in this study, and introduces SVM algorithm and ELM algorithm, and their performances were compared and verified through relevant experiments. It can be found from the results that for the dual branch CNN network structure, the joint classification of hyperspectral data and lidar data can achieve higher classification accuracy. For example, on different data sets, the global classification accuracy of the joint classification method reaches 98.46%. At the same time, experiments with the SVM model and the ELM model are carried out on different training sets. The DB-CNN model has the highest training accuracy and the fastest speed in training and testing. Besides, the test error of the DB-CNN model is the lowest at about 0.026, which is 0.037 lower than that of the ELM model and 0.056 lower than that of the SVM model. And the AUC value corresponding to the ROC curve of its model is about 0.922, which is higher than the other two models. It can obviously improve the effect of multi-source remote sensing data fusion and classification.