Identification of Aerial Image Land Use using Fused Thepade SBTC and Adaptive Thresholding with MachineLearning Ensemble

Land usage mining determines how a specific piece of land is used. This particular key can be used to identify different types of land utilization in aerial photos. This paper proposes the fusion of features related to Thepade SBTC and adaptivethresholding based features of aerial images for improved land use identification using machine learning. The Thepade SBTC (TSBTC) is investigated for land use identification from aerial images with fourteen different variations, including TSBTC-2ary, TSBTC-3ary, TSBTC-4ary, TSBTC-5ary, TSBTC-6ary, TSBTC-7ary, TSBTC-10ary, TSBTC-11ary, TSBTC-12ary, TSBTC-13ary, and TSBTC-14ary. Experiments are performed on a dataset known as the UCMerced Land Usage Dataset which has 2100 images representing 21 unique specimens of land usage. In the Thepade SBTC-14ary global feature extraction method, the Mathew Coefficient Correlation (MCC) and the F-measure showed improved performance. The proposed Fusion of Thepade SBTC and adaptive thresholding-based features have improved land use identification over consideration of individual features. Moreover, in addition to the nine separate machine learning algorithms, the suggested land use identification method also uses performance evaluation of MachineLearning algorithms. The majority voting-based ensembles “RandomForest+IBK+SMO” and “Simple Logistic+RandomForest+IBK” have led to more precise land use identification.