Integration of Object-Based Image Analysis and Convolutional Neural Network for the Classification of High-Resolution Satellite Image: A Comparative Assessment

During the past decade, deep learning-based classification methods (e.g., convolutional neural networks—CNN) have demonstrated great success in a variety of vision tasks, including satellite image classification. Deep learning methods, on other hand, do not preserve precise edges targets interest and extract geometric features such as shape area. Previous research has attempted to address issues by combining with object-based analysis (OBIA). Nonetheless, question how integrate those into single framework way that benefits each method complement remains. To end, this study compared four integration frameworks terms accuracy, namely OBIA artificial network (OBIA ANN), feature fusion, decision patch filtering, according results. Patch filtering achieved 0.917 OA, whereas fusion 0.862 OA 0.860 respectively. The CNN can improve accuracy; however, plays significant role this. Future should focus optimizing existing architecture, well investigate models use outputs for extraction remotely sensed images.