Today’s guest blog, written by Australian participant and trainer Ian McCan, is related to the use of 3D photogrammetry software and techniques. These techniques create a 3D image of an object by combining a number of photographs through a software programme (see also the blogs of the RCE dive team on the Oostvoornse meer: http://wp.me/p4KclF-4v) . This 3D object can then be rotated and send to people around the world as a means for research, for promotion purposes, or as a way to get the public involved. Ian McCan will now tell you more about the 3D photogrammetry project that he is involved with.
The 3DMAPPR (3D Maritime Archaeology Project – Perth Region) dates back to April 2014 and is a community-based project partly funded by an Austral-Asian Institute for Maritime Archaeology (AIMA) Scholarship in order to financially support the first stage of the program of shipwrecks site documentation, visualization and management. The main focuses of this starting step are:
(1) the development of a low-cost photogrammetry package intended to facilitate the (3D) recording of underwater cultural heritage in the Perth region
(2) the training of community members in underwater photogrammetric recording and image processing techniques.
The longer-term objectives for this project are for the hardware/software package and 3D imagery results to be used as a solution for low-budget (archaeology)groups or volunteer organizations as a management tool for the continual monitoring of endangered and important archaeological and non-archaeological sites.
This would also provide the ground work for the future use of augmented reality technologies as part of new virtual shipwreck trail visualization and for the use of 3D printing technologies within museum displays. This project is therefore not only focused on its application in archaeology as it is conducted today, but also aims to help create the future of archaeology.
Despite a comprehensive knowledge of maritime sites in the Perth Metropolitan area, the current status and condition of many sites remains somewhat uncertain due to a lack of up to date management and monitoring. This is associated with several additional factors, including a scarcity of detailed documentation for many local sites; a lack of any facility or tool to monitor the causes, nature, and scale of changes to local sites and their immediate environments; and a reliance on the involvement of (often) amateurs and hobbyists, with accompanying pressure on time, money, and expertise. Given these issues and the considerable constraints and limitations inherent in traditional (manual) survey and recording techniques, there was a clear need for alternative approaches to be adopted and implemented if on-going management of the sites in the Perth area and beyond is to be both timely and effective. Necessarily, any such approach needs to meet several criteria in order to make it fit for purpose. This includes time and cost effectiveness; ability to utilize off-the-shelf hardware and software systems; capability of operating with minimal user intervention; and accuracy and repeatability.
The technical details
A detailed literature review and preliminary land-based and underwater testing indicates that multi-image 3D photogrammetry best meets the above requirements. Multi-image 3D photogrammetry (MIP) is a term that describes the use of large 2D image datasets to reconstruct the 3D geometry of an object or scene using Structure from Motion (SFM) and Dense Multi-View 3D Reconstruction (DMVR) techniques. While the use of photogrammetric techniques has a long history in the context of land-based natural and cultural heritage documentation, its wider adoption and adaptation to underwater conditions has been considerably delayed owing to a number of technical and practical constraints and high technical overheads.
While several of these issues – particularly those related to underwater conditions (e.g. water turbidity, poor visibility, light attenuation and refraction) – remain, the advances in low-cost computing, digital imaging and software design have facilitated the development of what are effectively multi-image photogrammetry solutions. These solutions offer considerable advantages over traditional techniques, including rapidity, objectivity and relative simplicity of implementation; suitability for capture of large and complex objects; high potential accuracy; ability to use inexpensive and highly portable equipment (such as GoPro cameras); captured images contain all data required to facilitate 3D reconstruction; and the ability to utilize legacy data, thereby facilitating reconstruction and comparison of data from successive surveys and/or archive sources.
There are a considerable number of open-source (the Bundler + PMVS2 + CMVS assembly), web-based (123D Catch, Hyper3D/Cubify3D) and stand-alone (Photoscan, Photomodeller Scanner) multi-image photogrammetry solutions available at the moment. Each of these solutions offers varying degrees of user input and control over the resulting dataset. However, for the purposes of the project, it has been decided to employ Photoscan Pro, due to it being somewhat of a de facto standard in the field of archaeological photogrammetric documentation, being employed in a number of terrestrial and (increasingly) underwater scenarios. Unlike competing packages, Photoscan Pro represents a unified solution that incorporates not only the standard image acquisition, image rectification/alignment, and geometry extraction processing pipeline, but also has geo-registration capabilities and the option to output digital elevation models (DEM) and orthophotography (Geometrically corrected aerial photographs). These functions make it a versatile package suited for the many different archaeological needs.