Abstract: This paper integrates 3D discrete laser points and digital images acquired by close range laser scanner，non-contact digital camera，and rotation table to reconstruct 3D complex objects.According to the correspondent relations between 3D discrete laser points and the synchronous digital images，a new methodology of obtaining initial values for registration of two sets of 3D points is presented.All the range views are consolidated into one common reference frame under the constraints of the whole block adjustment.Based on the generated geometric model，this paper also studies how to realize texture reconstruction by considering the locations of objects and all the viewpoints.A small-sized cultural relic is utilized to testify the effectiveness and the feasibility of the proposed methodologies.High geometric accuracy and visual effects are achieved in this research.

Key words: laser techniques, point cloud registration, geometric reconstruction, texture reconstruction

摘要： 针对复杂目标的重建，利用近景激光扫描仪和机控的旋转平台获取目标的激光点云与数字影像两种不同的数据源，提出了一种大旋转角度的不同视点激光点云的配准算法，并实现了多视点云的无缝配准，建立目标完整的三维几何模型，对实验结果进行了精度评定。在此基础上，利用序列数字影像实现目标的纹理恢复，使得重建的目标既能够表现出精确的几何特征，又能够表现出丰富的纹理特征。实验证明，复杂目标重建的几何精度和纹理效果与实际量测结果和人的主观视觉感知具有良好的一致性。

关键词: 激光技术, 点云配准, 几何重建, 纹理重建