Abstract:We designed a view-dependent chunked terrain scheduling framework, and proposed an entropy-based method to fill the gap, which caused by different terrain LOD (Level of Detail) when the large-scale chunked terrain was loading and rendering. The method converted the 3D scene rendering result into image space, and distinguished the gap by the different texture color characteristic around the gap, then traversed the whole image space to find a best match using a fixed neighborhood, and confirmed the pixel"s color in the gap according the best match. During the traversal process, the texture color entropy was used as weight to filter pixels that had low possibility to become a candidate best match pixel to reduce the computational complexity, and GPU (Graphic Processing Unit) was also used to accelerate the computation. The experimental results illustrate that the method fills the gap smoothly when the user walkthrough the large-scale chunked terrain scene in real-time.