基于区域调制度和最大化的大尺寸物体三维重构  

A Three-Dimensional Reconstructed Method of Big Scale Objects Based on the Maximum of an Area Modulation Sum

作　　者：程旭升[1] 曹益平[1] 何宇航[1] 

机构地区：[1]四川大学光电科学技术系,四川成都610064

出　　处：《中国激光》2010年第3期784-788,共5页Chinese Journal of Lasers

基　　金：国家863计划(2007AA01Z333);四川省科技攻关项目(2006Z01-028);四川省学术和技术带头人培养基金(07GRC-01)资助课题

摘　　要：提出了一种对大尺寸物体进行三维重构的方法。该方法运用相位测量轮廓术(PMP)测得子区域的三维面形,采用区域调制度和最大化的思想进行高度相关匹配,通过拼接获得了整个物体的三维面形。该方法与运用所有高度数据进行匹配相比,减少了运算时间,提高了拼接精度。利用高度匹配得到的重叠区域位置,制作调制度二值化模板将物体从背景和阴影中提取出来,对子物体图像进行了灰度图像拼接。结果表明,三维重构的重复精度优于20μm,拼接精度优于100μm,为大尺寸物体的三维重构提供了一种有效方法。A novel and high precise three-dimensional（3D） reconstructed method of big scale objects is proposed.The sub-shapes of the experimental object are measured with phase measuring profilometry（PMP）,height matching is carried out in the modulation maximum area with an algorithm due to maximum correlativity.The method saves matching time by computer compared to using the whole height data of the sub-objects,and the mosaic precision is improved.Finally the whole shape of objects is obtained by the mosaic method.Binarization templates of modulation are generated in order to draw out the object from background and shadow,and image mosaic is carried out by using the overlapping position obtained from the above height matching.The results show that the repeatability precision of our method is superior to 20 μm,and the mosaic precision is beyond 100 μm,which is an effective method to reconstruct big scale objects.

关 键 词：图像处理 相位测量轮廓术 三维重构 相关识别 调制度 高度匹配 图像拼接 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]