Photogrammetry and the Future of Asset Digitization

Photogrammetry and the Future of Asset Digitization

2 Jul 2020 Written by Sharon

Photogrammetry is the technique of acquiring reliable physical measurements from photographic images. Coined by the Prussian architect Albrecht Meydenbauer, photogrammetry uses a variety of methods from many disciplines, such as optics and projective geometry, to create 2D or 3D models of objects based on their photographs. It has been used since the 19th century for many purposes; from creating topographic maps and elevation drawings, through architectural recordings, and, more recently, to engineering applications.

Stereophotogrammetry takes it one step further, by analyzing two or more photographs of the same object, and estimating three dimensional coordinates of points on the object’s surface. The small differences between photos taken from adjacent locations allow for depth analysis (also known as stereoscopy) and a better estimation of point positions in 3D space. Reliably applying stereoscopy to the photographs greatly depends on high contrast between the object and its background, which creates a well defined contour. The stereophotogrammetry process will be inefficient, therefore, when used on an object which blends with its surroundings, or when analyzing overly bright or dark photographs. These are key factors to consider when using drones to capture asset photographs. Stereophotogrammetry has become increasingly popular in asset management, and nowadays it and the broader photogrammetry term are being used interchangeably.

A successful photogrammetry process on a 3D object produces a “point cloud”: a vast collection of millions of points placed in 3D space that depict the captured object. The position of each point is set according to the photogrammetry tool’s calculation based on the photographs. The level of fit between estimation and reality depends on the systematic collection of the photographs, which takes into consideration the optimal angles and overlap between them.  Planning the images acquisition pattern, obtaining high quality photographs and applying sophisticated photogrammetry algorithms, are therefore instrumental for this process. The global advent of autonomous drone flights, specifically designed for the optimal collection of aerial photos, make these, once complicated and expensive steps, practical and affordable. Nowadays, comprehensive and detailed point clouds can be generated from a wide range of commercial and industrial assets. Using specialized software, point clouds can then be visualized on a computer, and measurements can be performed between any two points with a high level of reliability. 

Some photogrammetry tools go even further, by automatically drawing invisible lines between the points, based on the likelihood they are connected in reality. The resulting “mesh” of interconnected points and lines defines the 3D object’s surface. Cutouts are then taken from the original drone photos and “stitched” onto the mesh in the correct locations and orientations, generating a solid, photorealistic 3D model.  These models, commonly known as mesh models, usually produce impressive visual depictions of 3D objects that retain the textures and details of the original objects’ surfaces. However, due to estimations of line connections and photo cutouts “stitches”, mesh models are usually less reliable than their point clouds counterparts for very accurate (millimeters to centimeters) measurements.

Photogrammetry accelerates the access of information and shortens time to insights. These powerful analytical tools enable quick conclusions and decisions.  Both point clouds  and  mesh models eliminate the need for manual information collection, and enable robots to collect data, removing human errors in the process. The resulting 3D models allow better understanding of field assets and ensure accurate measurements for inspection.

vHive understands that photogrammetry is an indispensable tool in the era of asset digitization. We expect major technological improvements in this field in the coming years, such as industry-wide open standards, AI-assisted feature extraction, and more exciting innovations. Mainly, we foresee a shift from GIS-based photogrammetry models to IT oriented, automated continuous intelligence for the world’s industrial and commercial assets. 

Keeping its promise to provide the best drone hive software to its clients, vHive invests significant resources in the purchase, research and development of photogrammetry and 3D visualization technologies.

To learn more about vHive solution contact us at try@vhive.ai 

Related Articles