Multi-Temporal Land Cover Analysis of Vizag Using Hybrid OBIA

Land cover classification plays an important role in monitoring environmental changes and supporting urban planning. Traditional pixel-based classification methods usually have lower accuracy due to spectral similarities between land cover classes. In this paper, an Object based Image Analysis (OBIA) is adopted in which it improves the classification performance by grouping similar pixels into meaningful super-pixels. When Compared to Pixel-Based methods like Support Vector Machine (SVM), Spectral Angle Mapper (SAM), and Random Forest (RF), the OBIA approach significantly improved classification accuracy. To enrich segmentation, we further combined Simple Non-Iterative Clustering (SNIC) super segmentation and the Watershed Transform, forming a hybrid OBIA approach. Multi temporal land cover changes were being analyzed in Visakhapatnam over years 2017, 2010, 2023. Quantum Geographic Imaging System (QGIS) was used for data preprocessing, while Google Earth Engine (GEE) handled classification, enhancement and change detection. The model achieved high accuracy and results show major transformations whereas Future Predictions using the Markov Chain Model project further urban expansion and decrease in Agricultural land by 2029. This paper illustrates the effectiveness of hybrid OBIA in generating accurate land cover classification, contributing insights for urban planning and environmental management.