Scene Classification on Remote Sensing Images with Deep Convolutional Neural Network based on Modeling of Mayfly Optimization Algorithm

Abstract

Scene Classification (MFO-DCNNSC) technique on Remote Sensing Images. In the MFO-DCNNSC technique, the RS images undergo preprocessing using Wiener filtering (WF) to improve the image quality. To extract features, the MFO-DCNNSC technique applies the Inception-ResNetv2 model for learning the hierarchical depiction of the visual data. The MFO algorithm is useful for the appropriate range of the hyperparameters linked to the Inception-ResNetv2 method. With the great feature learning abilities of deep neural networks (DNNs), RSI scene classification compelled by deep learning (DL) has gained extraordinary attention and got important inventions.Scene classification is a vital study issue in RSI that has concerned numerous researchers presently. DL techniques are gaining a reputation in image feature analysis and reaching advanced performances in scene classification of RSI. This researchproposes a Mayfly Optimizer Algorithm with Deep Convolutional Neural Network. At last, the scene classification process has been executed by the usage of the deep belief network (DBN) model. To point out the improved performance of the MFO-DCNNSC approach, a huge assortment of experiments have been executed on the benchmark database. The obtained values pointed out that the MFO-DCNNSC technique outperforms the other models in terms of distinct measures.