Remote Monitoring Platform for Urban Road Projects Based on Ultra-High Resolution Remote Sensing Image Classification Algorithm

This study proposes a remote monitoring platform for urban road projects based on ultra-high resolution remote sensing image classification algorithms. The platform achieves efficient monitoring of urban road projects by combining deep learning technology and image analysis algorithms. As the first step of theoretical modeling, this study first analyzes the imaging characteristics of ultra-high resolution remote sensing images. At the same time, a road contour enhancement algorithm based on attribute morphological contours is proposed. This operation effectively solves the problem of blurred road boundaries and misclassification. Secondly, an efficient image classification algorithm based on the self-distillation model (SDM) is designed. The algorithm improves the classification accuracy of the entire model. It integrates a label encoder and a cross-attention module. Then, from the system design level, a comprehensive remote monitoring platform for urban road engineering is constructed. In additional, the overall architecture is demonstrated. The experimental results on the self-collected high-resolution remote sensing data-set show that the proposed image classification algorithm performs well in all 20 experiments. The classification accuracy of the proposed algorithm is between 99.74% and 99.87%. This result is significantly better than the traditional CNN, FCM and K-means algorithms.