Recurrence network modeling and analysis of spatial data.

摘要: Nonlinear dynamical systems exhibit complex recurrence behaviors. Recurrence plot is widely used to graphically represent the patterns of dynamics and further facilitates quantification patterns, namely, analysis. However, traditional methods tend be limited in their ability handle spatial data due high dimensionality geometric characteristics. Prior efforts have been made generalize a four-dimensional space for analysis, but this framework can only provide graphical visualization projected reduced-dimension (i.e., two- or three- dimensions). In paper, we propose new weighted network approach A model introduced data, which account both pixel intensities distance simultaneously. Note that each node represents location high-dimensional data. Network edges weights preserve structures patterns. representation shown an effective means complete picture Furthermore, leverage statistics characterize quantify properties features Experimental results simulation real-world case studies show generalized yields superior performance extraction salient systems.