Application of Historical Photos Photogrammetry in Documenting Destroyed Buildings (Case Study: Saadieh Historical Mansion)

Abstract
In some cases, photos are the only surviving documents of Iranian architectural heritage that can be valuable and unique documents in re-reading destroyed buildings. The reverse process of perspective drawing (graphical method) is often used to retrieve architectural documents, which is a complex and time-consuming method that involves errors due to ignoring lens or image distortions resulting from the reproduction of photographs in various media. The present study aims to suggest and apply a method based on which a single-photo rectified photogrammetry according to building geometry constraints could be applied to correct non-metric architectural photos that include a significant number of Iranian architectural visual documents. So the questions are what is the practical process of preparing metric documents through single-image reflection photogrammetry in historical photographs? What are the challenges, errors, and ways to achieve valid results in the case of Iranian architecture? And in practice, how can use this process in photogrammetry of the Saadi tomb as a case study to obtain reliable architectural documents? The research strategy is qualitative and case study and tries to present the process of converting photos into architectural documents based on revised image documents to be used in other research. Findings of the research suggest a flowchart including the steps of “pre-process”, “orientation” and “evaluation and analysis”, which have been adapted to work on non-metric historical photographs. The precision of photo rectification and possible errors in the correction process was categorized and described in three general sections, including errors of “digitization”, “reference point selection”, and “estimation and calculation of reference distances”. These errors could be minimized by repeating the process using various estimation methods, using parallel data in other available images, and using averaging as a way to find the most precise reference distances in complex situations. This method could be a reliable and fast way of producing architectural documents from images. The results of applying this method to the existing photos of Saadieh’s historical mansion showed the efficiency and accuracy of the process.
Keywords: Photogrammetry, Monoplatting, Photo Rectification, Documentation, Saadi Tomb.
Introduction
Documentation of historical monuments is the preparation of graphical and visual information that describes the physical arrangement, developments, and historical conditions of the building with accurate drawings. However, when a monument had been destroyed or access to it is not possible, the preparation of architectural documents from them requires methods other than the on-site survey. This is especially important in historical studies, which can be considered as various hypotheses about the building or as a basis for deeper analysis of the monument. Reading historical photographs from the perspective of architectural documentation requires methods that are sufficiently accurate and precise. Architectural photogrammetry, especially in the last two decades, with its growth in digital technologies and processing software, has made reliable tools available to researchers that are evolving every day. The purpose of this article is to develop a practical process that can be used to rectify historical photographs of Iranian architecture and applied to buildings that have been destroyed or severely eroded, by extracting metric data from them and producing the necessary architectural documents. The methodology of this process and its levels and stages can be used as a way to deal with other existing visual documents of Iranian architecture and to help extract comparable architectural documents from photographs. To know the way of working and the challenges of the researchers as much as possible in advance and find a solution, the former mansion of Saadieh in Shiraz, which was built in the time of Zandieh, has been considered as a case study. 
The main questions of this research are as follows: What is the practical process of metric reading through photogrammetry, single-reflection in historical photographs? What are the challenges and their validation in Iranian architectural monuments? And in practice, how can we use this process in the photogrammetry of the Saadieh mansion (in terms of design and geometry of the facade) to obtain reliable architectural information, and what will be the validation of the findings? The strategy of this research is qualitative. First, the basic concepts and key steps of photogrammetry are presented. Then the steps of rectifying the historical photo are introduced and finally, the photogrammetry of the selected photos of the Saadieh mansion is given. In this research, Faro-As-Built version 2019 has been used, which is one of the most practical and accurate photogrammetric software in architecture and documentation.
Identified Traces
In this article, a flowchart for photogrammetry of historical photographs is proposed. These steps are similar to almost all methods and systems, regardless of the type of software used. The proposed flowchart consists of three logical steps. In step (a) or the pre-process, the appropriate photo or photos should be selected according to the criteria set out earlier. Then if they are not digital, they should be scanned. In the next step (b), the orientation process is performed. The lens distortion must first be corrected. Photogrammetry software has its instructions for removing camera lens distortion, in which the user must perform a set of actions on the image according to the standards of any software. But in general, all of them use a simple method to eliminate lens distortion in non-metric cameras, and that is the comparison of horizontal and vertical lines parallel to the building that must be defined by the user in the software environment. The accuracy and frequency of the lines introduced by the user can be important in the quality of the final result. Selecting reference points and controlling is another step in the justification section, which is very important and is the key step after correcting the lens distortion or internal camera alignment. The final step (c) is the measurement and evaluation, which requires the addition of a correction grid to the image. This network must be adapted to the perspective of the building and adjusted based on the reference points and the desired correction page. The reference point data must then be provided to the software for correction.
Conclusion
Today, with the development of digital technologies, new tools are provided to rectify historical photographs that can provide researchers with a more accurate understanding of architectural heritage. In this research, a three-step method including “pre-process”, “orientation” and “measurement and analysis” for photogrammetry of historical non-metric photographs was proposed. Errors in the process of photogrammetry Historical photographs can be generally classified into three categories, including “digitization errors”, “reference point selection” and “reference size estimation errors”. The first error could be considered “tool-oriented” and the other two should be considered “user-oriented”.
Although the errors resulting from “reference point selection” can be greatly remedied by increasing the skill and experience of the user and repeating the photogrammetric process by defining various reference points, errors in “reference point selection” and “reference size estimation” should be considered the most fundamental flaw in the course of work and results. This requires skill, knowledge, and creativity at the same time; In general, identifying the geometry and proportions in Iranian architecture, familiarity with styles and methods in building facades designs and a correct understanding of the spatial structure of the building along with reviewing existing sources and historical documents can be considered basic skills to solve these errors.