Enhancing Statistical-Based Remote Sensing Image Classification Algorithms: An Optimization Study

This study presents an optimization approach for statistically based remote sensing image classification algorithms. Traditional image classification methods suffer from subjectivity and inefficiency due to manual annotation. Taking advantage of advances in artificial intelligence and computer vision, this study focuses on content-based image classification methods. Specifically, the study focuses on optimizing statistically based algorithms for classifying remote sensing images, which are known for their complexity. The methodology includes efficient image feature extraction using visual saliency features based on regional covariance, followed by an optimized statistical-based classification algorithm using extended sparse coding and an empty spectrum dictionary model. Experimental evaluation conducted on remotely sensed images of Hong Kong demonstrates the superior performance of the proposed method compared to existing approaches, with an average accuracy rate of 99.41%. A comprehensive error analysis highlights the robustness and consistency of the proposed approach, confirming its effectiveness in addressing challenges of remote sensing image classification.