基于深度图像的多特征孔洞填补方法研究  

Multi-feature Hole Filling Based on Range Image

作　　者：赵翠莲[1] 周凯[1] 范志坚[1] 陈昱[1] 

机构地区：[1]上海大学CIMS和机器人中心,上海200072

出　　处：《机械设计与研究》2008年第3期15-17,25,共4页Machine Design And Research

基　　金：上海市教育基金资助项目(06AZ023)

摘　　要：为填补深度图像中包含多个特征的孔洞,提出了一种基于模板匹配的孔洞填补方法。根据法矢进行区域特征分割,得到单一特征点云,运用切片法从单一特征点云中得到特征线点云。对特征线点云中值滤波后拟合生成特征线,并对特征线编码。同时建立模板编码库,采用模板匹配法匹配模板图和搜索图中各特征线子图,找出最优模板图进行孔洞填补。该方法应用在船舱槽型舱壁的重构中,能有效地实现多特征孔洞填补。To repair the hole inthe range image which includes several features, a method of hole-filling based on template matching was proposed. The whole image was segmented according to normal and point clouds of single feature were obtained. After point clouds of the feature lines of single feature were obtained using slicing method, the feature lines were fitted by median filter and coded. Code database of template was constructed. When the optimum template image was found based on template-matching, the multi-feature hole can be filled. The method is applied to the reconstruction of corrugated bulkhead, which demonstrates that it is obviously helpful for multi-feature hole filling.

关 键 词：深度图像 多特征孔洞填补 编码 模板匹配 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]