Detection of land cover usage from optimized learnable parameter artificial neural network (OLPANN) using multispectral images

Detection of land cover usage from optimized learnable parameter artificial neural network (OLPANN) using multispectral images

연구 분야: Networking

학회: Multimedia Tools and Applications

Land cover classification is a vital task in remote sensing to emerging demands in global and echo-friendly environmental applications. This type of analysis will empower with different application such as agricultural phenomena, climate change identification, natural land covers conversion and etc. We proposed the Optimized Learnable Parameter Artificial Neural Network (OLPANN) model for identifying land cover delineation boundaries. The OLPANN model will examine the potentials of spectral, spatial, and texture indices information from the multispectral data using Pearson correlation measures. The extracted deep features from the original dataset will increase robustness of the Artificial neural network model. To provide a faster and mutable classified result, the extracted features are trained by the ANN. The sentinel-2 dataset was used to analyse three different types of features for evaluating seven different land cover classes. McNemar’s test was carried out to evaluate the changes which endorsed OLPANN and (Optimized Extreme Gradient Boosting) OXGB is statistically significant. Friedman’s test demonstrated that variance of (Optimized Random Forest) ORF, (Optimized Support Vector Classifier) OSVM, (Optimized Decision Tree) ODT were significant at 0.01% level. The numerical outcomes obtained have established that OLPANN can achieve the highest accuracy of 94.07%. Hence OLPANN classifier can be recommended as the best candidature for its ideal measures for the salient characteristics considered for the study. This type of analysis will empower the government to identify the urban extension, damage delineation of natural land covers, legal boundaries for property assessment, and target detection like roads, bridges, and water or land surface interfaces.