基于统计模型的三维场景重建补洞算法

Hole Filling in Range Image Procesing Based on Statistical Model

提出了一种新的处理距离图像(即点云数据)算法。算法提取已经获得的表面数据的特性,计算各个被测的特征值,建立马尔可夫统计模型,描述整个表面的数学特征,进而利用最大似然概率预测丢失位置的数据值,得到被测物体表面的完整的数据描述。使用该算法恢复由于障碍物和表面不良反射特性产生的空洞,具有良好的视觉效果,与测量获得的表面数据具有统一的几何特征。

A set of data processing algorithms was developed for generating whole point cloud description of building from a piece of range image,obtained by a laser scanner placed on one fixed spot.The range image with some data lost,which is made up of unorganized points,was taken as input,and then a whole set of points to describe the surface was outputted.Markovs statistical model based on the repeated geometry pattern of the surface was created to predict the position of the lost points and fill the hole on the surfaces.The result of this filling hole algorithm has the similar visual effect with the surface data that got from the laser scanner.