在以前的基于目标空间划分的并行体数据绘制算法中，局部绘制和图象融合是两个串行的过程，在节点机的局部绘制阶段几乎没有数据通讯，但在数据融合阶段数据通讯量非常大，出现总线争用甚至通讯阻塞，而且在这个阶段有非常大的同步开销。本文利用流水线结构，让局部体数据绘制和图象融合并行执行，很好地解决了上述缺点。并在一个基于微机的流水线结构上实现了一个新的基于目标空间划分的并行体数据绘制算法。

In distributed computing environment, an important factor which will affect the performance of parallel algorithm is communication bandwidth. This paper presents a parallel volume rendering algorithm using a shear-warp factorization of the viewing Transformation on the local area network of PCs. We fully take use of the overlap of communication and computing to overcome the bottleneck of communication. In many parallel volume rendering algorithm based on object partition, local volume rendering and image composition are divided into two serial processes, in the period of local volume, the communication hardly happen, contrarily, in the period of image composition, the communication is very busy, even communication congest happen, furthermore there are a very big synchronism overhead in this period. The paper take use of Pipeline based on PC, local volume rendering and image composition are executed concurrently, the draw backs above are solved well. Our experiment which is completing on the Pipeline including 16 Pentium shows that communication do not affect the performance of algorithm, and the overheads are very little to rendering time. The paper provide a method for studying low-price, high-efficiency real-time volume rendering system.