结合可见性处理的顺序点模型树

将点模型的树节点按照各自的误差进行顺序排列，就形成了顺序点模型树，这样，就能方便地利用GPU(图形处理器)来高速绘制点模型，但这种方法缺乏对可见性的高效处理，不便于处理大型复杂的模型，因为这些模型中各个部分之间的遮挡关系复杂，且每次成像所需的树节点的误差跨度大，导致了大量不可见的树节点也要绘制，为此，文章提出一种结合可见性处理的顺序点模型树，即在根据误差顺序排列树节点时，融入一些可见性信息，这样，绘制时，不仅能保留顺序点模型树高效利用GPU的优点，还能避免处理大量的不可见但满足误差要求的树节点，实验表明，相比于顺序点模型树，新方法的绘制速度更快，加速率至少在20％，特别是在处理大规模复杂模型时的加速率更高，能达67％以上。

Sequential Point Trees with Visibility Processed Efficiently

A sequential point tree is constructed from a hierarchical point tree by rearranging the tree nodes in a sequence according to their respective geometric errors, which benefits using GPU for fast point rendering. However, it lacks efficient measures for visibility determination so that many points that satisfy the error requirements for rendering but invisible may be rendered. This will seriously lower the efficiency in rendering the large and complex model because the points satisfying the error requirements but invisible will be in a large quantity due to the complicated occlusion relations between different parts of the model and the wide error range permitted in rendering the model. As for this, a new method is given in this paper to construct the sequential point tree with some visibility information considered in rearranging the tree nodes and integrated in the sequential points. Therefore, in rendering, not only the GPU can be conveniently employed for acceleration, but also a lot of points that satisfy the error requirements but invisible can be skipped for acceleration. Experimental results show the new method is at least 20% faster than the existing sequential point tree, and it is much faster in treating the large and complex models, more than 67%.