在曲面重构中，由于实际的曲面模型往往含有多个曲面几何特征，即由多张曲面组成，如果对使用激光法测量的“点云”数据直接进行拟合，将会造成曲面模型的数学表示和拟合算法处理的难度加大，甚至无法用较简单的数学表达式描述曲面模型，因此针对该问题，提出了一种基于数据点曲率变化的区域分割方法，即先对每一条扫描线上的数据点求取曲率值，然后将其中曲率值变化较大的点提取出来作为边界点，当边界确定后，再将云点数据分割成多个区域，由于每个区域一般具有较简单的几何特征，因此可用简单的数学模型来描述，并可重构单张曲面。该算法不仅原理简单、易于理解和编程，而且能提高曲面模型重构效率。

In surface reconstruction, the actual surface model generally contains several surface geometrical features. Namely it consists of multi-surface patches. If directly fitting surface with these point cloud data obtained from a laser scanner, the surface model presentation and the surface fitting are hard to do. Even it is difficult to use a simple formula to describe the surface model. In this paper, a method for region segmentation algorithm according to the curvature variation of data points is proposed. The operational principle is to calculate the curvature values to data points on each scanning line. The points with bigger curvature variation are picked up as boundary points. After borderline are determined, point cloud data are divided into several regions. Each region has simple geometrical features, and can be described in simple mathematical model, and then a single surface is reconstructal. This algorithm is simple, and can be easily understood and programmed. Efficiency of surface model reconstruction is enhanced.