采用平面特征搜索的深度图像坐标对准  

Normalization of Range Image Based on Planar Feature Searching

作　　者：陈昱[1] 赵翠莲[1] 周凯[1] 范志坚[1] 

机构地区：[1]上海大学机器人与CIMS中心,上海200072

出　　处：《工程图学学报》2009年第4期113-117,共5页Journal of Engineering Graphics

基　　金：上海市教委基金资助项目(06AZ023)

摘　　要：提出了一种包含平面特征的激光扫描深度图像行之有效的坐标自动调整方法。通过移动包围盒的方法在数据点集内自动搜索并判定特定平面特征作为基准;计算出平面特征的平均法向量,将点集的坐标向基准平面的法向对准;最后通过缩小包围盒范围并重复搜索,使点集的方向规范化。此方法在液货舱激光扫描数据上进行了应用,数据经过对准处理后,大大方便了后续的投影、分割等模型重建工作。An automatic normalization method for range images contained planar features is proposed. Laser-scanned range images are transformed into canonical coordinate frame by the following steps： the first is to use moving box to search plan automatically; then the data points are rotated and aligned to the direction of the plan; by reducing the size of the searching box and repeating the search-rotate process several times, a desired point data set is obtained. This method has been applied to scanned data of ship cabin. The results of normalization are as expected and thus simplify the following works such as projection, segmentation, etc.

关 键 词：计算机应用 坐标对准 特征搜索 深度图像 逆向工程 

分 类 号：TP391.72[自动化与计算机技术—计算机应用技术]