Remote Sensing Change Detection Based on Unsupervised Multi-Attention Slow Feature Analysis

With the development of big data, analyzing environmental benefits transportation systems by artificial intelligence has become a hot issue in recent years. The ground traffic changes can be overlooked from high-altitude perspective, using technology multi-temporal remote sensing change detection. We proposed novel unsupervised algorithm combining image transformation and deep learning method. new for images is named multi-attention slow feature analysis (ASFA). In this model, three parts perform different functions respectively. first part records to K-BoVW classify categories objects as channel parameter. second residual convolution with multiple attention mechanisms including temporal, spatial, attention. Feature extraction updating are completed at link. Finally, we put updated features highlight variant components which want then generate map visually. Experiments on very high-resolution datasets verified that ASFA better performance than four basic detection algorithms an improved SFA algorithm. More importantly, model works well road helps us analyze changes.