Focusing on the complex data structure of digital elevation model (DEM) and the poor rendering speed in real terrain visualization

we propose a fast rendering method based on adaptive multi-feature fusion for realistic terrains. To introduce the entropy of terrain elevation

real DEM elevation data can be extracted for generating an overall framework. According to the random midpoint displacement fractal algorithm and optimized fractal parameters

the high frequency detail can be increased. To calculate the distance threshold between the viewpoint and the terrain

this corresponds to the level of details (LOD)

so that it can achieve adaptive scheduling. Additionally

by using the uncertainty determinant factor

the characteristics of the terrain profile can be updated. Finally

the algorithm is carried out using parallel processing

taking full advantage of the graphic processing unit (GPU) for accelerating the terrain rendering. The experimental results show that the generated terrain has higher fidelity and better real-time capabilily.