Purpose

This paper aims to propose a system to acquire images automatically for digital reconstruction of heritage artifacts using a six-degree of freedom industrial manipulator.

Design/methodology/approach

A virtual environment is created using Robot Studio® software to integrate the trajectory and differential motion of the robot manipulator and the motion of camera while acquiring images. A new area similarity matrix method is proposed to reduce the number of images required for digital reconstruction using Autodesk Recap® software. Real-time experiments have been performed using objects such as minion, ultimaker robot and cube. Evaluation of the digital reconstruction is conducted using the contour area matching method.

Findings

The number of images required for reconstruction based on area similarity matrix method is reduced to 63 per cent when compared with the random selection method. Quality parameters such as surface area, volume, number of defect holes, vertices and faces are enhanced for the proposed method.

Research limitations/implications

Digital reconstruction of large-sized heritage artifacts cannot be performed in this setup. But this can be overcome by fixing the manipulator on a mobile platform or overhead crane. This paper does not discuss the reconstruction of partially damaged heritage artifacts, which could be accomplished based on deep learning techniques.

Practical implications

Using this approach, off-the-shelf heritage artifacts and large-scale objects can be reconstructed digitally with a minimum number of images and without compromising the quality of original models.

Originality/value

To the best of the authors’ knowledge, area similarity-based approach in 3D digital reconstruction by coupling the kinematics of an industrial manipulator and camera is proposed for the first time. A fully automated digital reconstruction technology to preserve valuable heritage artifacts has been developed. It also highlights the space constraints of the industrial manipulator in digital reconstruction.

中文翻译：

---

机器人协助的3D数字重建文物：区域相似度方法

目的

本文旨在提出一种使用六自由度工业机械手自动获取图像以进行文物人工重建的系统。

设计/方法/方法

使用RobotStudio®软件创建虚拟环境，以在获取图像时集成机器人操纵器的轨迹和差动运动以及相机的运动。提出了一种新的区域相似度矩阵方法，以减少使用AutodeskRecap®软件进行数字重建所需的图像数量。实时实验已使用小兵，超人机器人和立方体等物体进行。使用轮廓区域匹配方法进行数字重建的评估。

发现

与随机选择方法相比，基于区域相似度矩阵方法重建所需的图像数量减少到63％。该方法提高了质量参数，如表面积，体积，缺陷孔的数量，顶点和面。

研究局限/意义

在此设置中无法执行大型遗产文物的数字重建。但这可以通过将机械手固定在移动平台或高架起重机上来克服。本文没有讨论部分受损的文物的重建，这可以基于深度学习技术来完成。

实际影响

使用这种方法，可以用最少数量的图像以数字方式重建现成的文物和大型对象，而不会影响原始模型的质量。

创意/价值

据作者所知，首次提出了通过将工业机械手和相机的运动学结合起来的3D数字重建中基于区域相似度的方法。已经开发出一种用于保存有价值的文物的全自动数字重建技术。它还强调了数字重构中工业机械手的空间限制。