Multiple-point Geostatistical Simulation for Modeling Positional Uncertainty in Remotely Sensed Images
In order to model and analyze the positional uncertainty of remotely sensed images,a multiple-point geostatistical simulation method considering the trend surface is proposed. Firstly,the trend surface of image positional errors in the study area is fitted. Then TI of residuals is generated by using the residual of study area,and direct sampling (DS) simulation is carried out for the study area based on TI of residuals,thus using random simulation to achieve positional uncertainty modeling. Experimental results show that com-pared with conditional simulation,unconditional simulation is more suitable for random simulation with positional uncertainty;when the positional error has an obvious overall trend,the unconditional simulation method considering trend surface has a better simulation effect. The proposed method is also applicable to the uncertainty modeling of other geospatial variables with overall trends.
万方数据
维普
