A Novel Dual Temporal Channel Convolutional Network with Exponential Distribution Optimizer for Satellite Image Classification with Google Earth Engine

Classification of satellite images using Google Earth Engine entails identifying features on the satellite images to classify the land cover in a particular region through the use of deep learning algorithms. Such a process contributes to monitoring the environment and geographical distribution of resources, planning of cities, and management of resources. However, in many existing approaches, some challenges are worth noting. To address these challenges, this study introduces a new approach, the Dual Temporal Channel Convolutional Network with Exponential Distribution Optimizer (ANDualTCCNet+EDO) for satellite image classification using Google Earth Engine. This approach uses satellite imagery from the Landsat 8 Surface Reflectance dataset, with the input images preprocessed through the Adaptive Tri-Plateau Limit Tri-Histogram Equalization Algorithm (ATri-PLTri-HEA) technique. Following preprocessing, features are extracted from the images using the DenseNet201 method. Then, classification is carried out by A Novel Dual Temporal Channel Convolutional Network (ANDualTCCNet), optimized by the Exponential Distribution Optimizer (EDO) for precise satellite image classification in Google Earth Engine (GEE). The ANDualTCCNet+EDO method is implemented in Python. The proposed method was evaluated using the Landsat 8 Surface Reflectance dataset, achieving a high accuracy of 99.8%, surpassing existing methods. The results presented in this paper show that the use of this method allows for the accurate classification of satellite images in Google Earth Engine. The outcomes of the proposed method improve the method of satellite image classification in Google Earth Engine. This model captures both short- and long-duration temporal characteristics, therefore enhancing spatial-temporal feature acquisition compared to the previous method used in the classification of land cover, especially for temporal dynamics of land covers with different seasonal changes.