Photogrammetry is the New Point Cloud (and Vice Versa)

Its quite apparent, and probably obvious to everyone who reads LiDAR News, that point cloud capture, visualization and analysis technologies are becoming ubiquitous across every market sector that deals with real-world objects. Whether youre a designer of manufactured parts or a civil engineer, you have probably used or considered using point clouds at some point in time (no pun intended) to capture an accurate representation of something youre working with, in, or on. With this ubiquity, point cloud scanners are becoming cheaper and cheaper, which of course spawns further use and growth of these innovative technologies.

Point clouds are perfectly suited for extremely accurate scans of large areas, during the night, and through clouds when 3D survey accuracy is needed. However, another technology trend that seems to be growing at perhaps a faster rate, is the use of photogrammetry with and without a point cloud as a way to perform preliminary surveys and measurements (as well as a bunch of other uses) when traditional laser-scanned point cloud accuracy is not needed. And, while point cloud scanners will very likely make it to our mobile phones someday, cameras are already everywhere. Moreover, stereo cameras are already standard hardware on many devices.

Stereo cameras and photos alone do not an accurate 3D scan make. In order to create accurate 3D representations of real world objects you need purpose built software that understands how to blend stereo-pair images into 3D scenes. While the images themselves might be cheap to produce, its the software thats special. There are a few great examples of companies and software on the market that are changing the game when it comes to 3D photogrammetry. earthmine, a company out of Berkeley, California, employs Mars Rover technology (they hold an exclusive license), custom cameras, accurate GPS and IMUs, as well as special cloud-driven software to produce what are effectively 32 mega pixel panoramic images with where every pixel describes depth. To the end user earthmine images are viewed over the web or within desktop software much like Google Street View.

But earthmine provides so much more. Via the images alone, users can perform vertical, horizontal, area and 3D measurements with sub-meter accuracy globally. While this is not nearly as accurate as a traditional point cloud, for many applications, like asset management (GIS), what earthmine provides is more than accurate. And the cost and process end-to-end, from capture to actually using the imagery, can be a fraction of what it costs in money and time it would take to capture and process large linear features, like an entire citys street network, using traditional point cloud technologies. Figure 1. earthmine extension for AutoCAD Map 3D user interface and measurement tools >>

Pictometry is another good example of a company that employs photogrammetry and software to produce accurate 3D imagery and associated tools. Whereas earthmine cameras are typically mounted on a street vehicle and used to capture streets and highways, Pictometry owns and operates a large fleet of planes which they affix with high-resolution aerial cameras. In Pictometrys case, instead of using stereo cameras to capture 3D information, they use a very tight grid pattern for their flight path (as well as accurate GPS and IMUs), ensuring that they capture all four sides of any object on the ground at oblique angles. Using proprietary software to match up the images, Pictometry can produce accurate 3D models of entire cities. And, like earthmine, they provide software that enables end-users to perform horizontal, vertical and 3D measurements using only the imagery.

Offering yet another way to utilize standard imagery to create 3D models is Autodesk 123D Catch. In the case of 123D Catch, users (any user) take multiple photos of any object using any digital camera, including smart phone cameras at a bunch of different angles (usually every five degrees) and import the images into the 123D Catch software. 123D Catch then sends the photos to The Cloud for processing and returns accurate 3D models of the object. These models include all of the color and detail of the original photos, as well as extremely tight 3D meshes that can be used to further manipulate the models. << Figure 2. Pictometry oblique image

Given that 123D Catch is free, Autodesk is in essence democratizing (and consumer-izing) the ability to capture scans of real-world objects in 3D with nothing more than a camera or smart phone and simple to use software. Thats not to say photogrammetry and point clouds technologies dont mix, in fact quite the opposite is true. Many, if not most, of the top point cloud scanner vendors, including TopCon and Faro, already include really innovative ways, via hardware and software, to mash-up photos and point clouds in order to produce stunning 3D scenes that also include LOTS of points. Whats more, the photos in this case can be used to augment the point cloud where you dont have perfect coverage. Figure 3. 123D Catch software >>

Take for example a point cloud scan of a room or a building, where the goal is to produce a near perfectly accurate 3D model. In both cases the point cloud will be very effective at capturing most of the detail required for the 3D model, except when it comes to the corners (interior or posterior). For corners, if you dont have a point at the very base or apex of the corner, there is a potential that your model will be off. If your goal is accuracy, this wont work. However, if you bring photogrammetry into the picture (again, no pun intended) overlaying points on top of the photos there is a good chance that youll have tools available to see the corners and potentially add points where needed, thereby creating a more accurate model.

In a world that is seeing 3D capture technologies growing at a rapid rate, it is important to remember that there is a time and place for every technology and that sometimes the newest technology doesnt necessarily provide the best fit for every problem. Moreover, sometimes the best fit for a project comes from blending multiple innovations into one solution, thereby addressing the widest array of possible use cases. In this way, you might even say photogrammetry is the new point cloud is the new photogrammetry.