集群技术下的实景三维建模

计算机集群技术的出现为倾斜摄影测量中实景三维自动化建模时遇到的计算资源匮乏问题提供了解决途径。本文论述了集群在实景三维建模中的两种应用模式：基于物理并行计算的集群和基于逻辑并行计算的集群；阐述了在逻辑并行计算的集群模式下ContextCapture软件进行实景三维建模的流程；提出了该集群模式下子测区划分的5个原则：起降点原则、飞行高度原则、建筑物密度/植被密度原则、最大计算量原则、航线角度原则，并有效控制了划子区后的精度问题。最后，本文利用实验室多台普通PC机实现了某测区实景三维集群建模，结果表明：该方法自动化建立模型精度满足测区需求，可有效加快自动化建模速度，降低生产成本。

Research on real 3D modeling supported by cluster technology

The emergence of computer cluster technology provides a solution to the shortage of computing resources encountered in 3D automatic modeling of tilted photogrammetry. This paper discusses two application patterns of cluster in 3D modeling:cluster based on physical parallel computing and cluster based on logic parallel computing. The process of 3D modeling of real scene in ContextCapture under the cluster mode of logical parallel computing is expounded. Five principles of subdivision in this cluster are proposed:the principle of the starting and landing points, the principle of flying height, the principle of building density/vegetation density, the principle of maximum calculation and the principle of line angle. And the accuracy of the sub region is effectively controlled. Finally, this paper uses a number of common PC machines in the laboratory to achieve a three-dimensional cluster modeling in a certain area. The results show that the method automatically establishes the model accuracy to meet the requirement of the survey area, and can effectively speed up the automation modeling speed and reduce the production cost.