中文摘要：

通过对风景园林规划合理性的三维图像分析，完成风景园林合理分布系统设计，能够有效提高园林景观规划的合理性。对风景园林规划分布合理性三维图像的分析，需要先对图像特征点检测，并对这些特征点进行聚类匹配，完成对风景园林分布的三维分析。传统方法先获取风景园林图像的特征点，计算风景园林场景的空间坐标，但忽略了对特征点的聚类匹配，导致分析结果误差较大。提出一种基于三维视觉的风景园林分布合理性三维图像分析方法。上述方法先基于特定视角下不同坐标系之间的关联性组建相机位姿矩阵，对风景园林图像进行SURF特征点检测并对上述特征点进行聚类匹配，根据配准的图像特征点为依据实现风景园林图像三维图像重建，以风景园林三维图像重建结果为依据完成对风景园林分布合理性三维图像分析。实验结果表明，所提方法风景园林三维图像重建精度较优。

英文摘要：

In this paper, we propose a three -dimensional （3D） image analysis method for distribution rationality of landscape architecture based on 3D vision. Firstly, based on relevance among different coordinate systems under specific angle of view, a position matrix of camera was built. Then, SURF feature point detection was carried out for image of landscape architecture and cluster matching was carried out for the feature point. According to registration feature point of image,3D image reconstruction of landscape architecture was achieved. Finally, according to the reconstruction result, the 3D image analysis was completed. The conclusion can be drawn from the experimental results that the method has optimal precision of 3D image reconstruction.