LatentMap: Effective auto-encoding of density maps for spatiotemporal data visualizations

Shiqi Jiang, Chenhui Li, Lei Wang, Yanpeng Hu, Changbo Wang

Abstract: Highlights•We propose a GAN-based method to analyze the latent space of density maps, and latent codes are available for compression to save transmission time. Our method can be applied to any density maps in the spatiotemporal visualization.•Based on latent codes, we calculate the dynamic changes of the density map by morphing, which can complete missing data in a spatiotemporal visualization.•We collect thousands of density maps in three ways and process them for training and testing. We demonstrate the effectiveness of our method in three forms of density maps such as spatial density maps and slices of the volume data.