基于八叉树空间分割的三维点云模型密写

摘要

Aiming at the information hiding of 3D point cloud model, this paper presents a spatial domain steganography algorithm based on octree space division. Preprocessed by Primary Component Analysis(PCA), the bounding box of the 3D point cloud model is established and subdivided into voxel based on octree, in which octree structure is used to record segmentation process. The hiding messages are then embedded in the subspace of the voxel by moving the point position. Experimental results show that the extraction process of the algorithm does not need original data model when extracting information, it has features of high capacity and low distortion, it is robust against affine transformations and vertex reordering attacks and also suitable for any meshes of 3D model.%针对三维点云模型的信息隐藏,提出一种基于八叉树空间分割的空域密写算法.对经过主成分分析后的三维点云模型建立包围盒,利用八叉树空间分割得到小体元并记录分割过程,通过顶点位移将信息嵌入到小体元内的不同空间位置.实验结果表明,该算法在提取信息时不需要原始模型数据,具有嵌入量高、失真度低的特点,能够抵抗旋转、平移、均匀缩放和顶点重排序攻击,适合于任意网格的三维模型信息隐藏.