为了更好地评价大口径反射光学元件在不同尺度下的起伏情况，提出了一种基于结构函数的子孔径拼接算法。首先，对于算法的基本原理与步骤进行了描述，从理论分析的角度对于计算误差的特性进行了分析；之后针对口径为1.23 m的大型反射镜面形数据，应用文中所提出的方法，利用结构函数进行子孔径拼接并对于其误差特性进行了检验，验证了所提出方法的可行性。最后，将该方法应用于30 m望远镜三镜(TMT M3)的面形仿真数据，得到了其在不同评价尺度下的起伏情况。文中的工作对于TMT M3 的最后完成有着重要的意义，并且对于与TMT M3 类似的大口径平面镜面形评价有着一定指导价值。

In order to specify the ripple of the large reflecting optical item in different scale, the algorithm based on the structure function was investigated. Firstly, the concrete steps of the algorithm were presented. Secondly, the error composition character was discussed in theory. Furthermore, the method proposed here was applied to the mirror figure data of a 1.23 meter mirror, making the conclusion that the simplification was correct and application was feasible. Lastly, the application to the thirty meter telescope tertiary mirror was processed, the ripple of the mirror figure in different levels was specified. This work will guide the way to the large flat mirror specification.