Abstract: Safety production supervision departments have strict regulations for the five-distance of the hazardous chemicals storage warehouse stacks (distance, pile wall distance, zenith distance, spacing and channel spacing), the dynamic supervision requirements can be achieved by describing the changes of the position of the stacks in the warehouse through the three-dimensional dynamic geometric modeling. Three-dimensional geometric dynamic modeling of the stacks based on image is proposed. In order to establish the corresponding relations between image points and objects, a five - element imaging model is constructed and two - point calibration method is proposed. On the basis of calibrating the parameters, the solution space of the coordinates of the point is obtained first, and then the real solution space is obtained by combining the constraint condition to realize the static  three - dimensional geometric reconstruction of the stacks. Then, the stacks are identified and reconstructed in the image sequences, and the dynamic geometric modeling is realized.

Key words: Five-distance, 3D geometry reconstruction, five-element imaging model, two-point calibration method, dynamic geometric modeling

摘要： 安全生产部门对危化品仓储堆垛五距（堆距、墙距、顶距、柱距和通道距）有严格的规定，通过对堆垛的三维动态几何建模描述其在仓库中位置的变化，可以达到动态的监管要求。提出了基于图像的堆垛三维几何动态建模，为建立像点与物点的对应关系，首先构建了五元组成像模型并提出了两点标定法，在标定好参数基础上首先求得物点坐标的解空间，再结合约束条件得到真实解空间，实现堆垛静态三维几何重建。然后对图像序列中堆垛箱体跟踪识别并三维几何重建，最终实现动态几何建模。

关键词: 五距, 三维几何重建, 五元组成像模型, 两点标定法, 动态几何建模