3D Scanning: Field Productivity Factors in Laser Scanning, Part 2
Professional Surveyor Magazine - February 2007
In many cases laser scanning can offer significant field productivity advantages over traditional surveying methods for a given topographic or as-built survey project. For professionals who are investigating laser scanning systems, understanding various field productivity factors for these systems can be very valuable.
Part 1 of this series (Professional Surveyor, January 2007) detailed the field productivity aspects of site reconnaissance, system portability, and the number of instrument setups as they apply to high-definition surveying. These three elements are all common to traditional surveying as well, although each has its particular twist for scanning. Part 2 of this 3-part series considers an aspect of field productivity that is unique to high-definition surveying: the number of target setups.
Number of Target Setups
Special scan targets can be used for accurate registration (or stitching together) of scans taken from multiple scanner positions and for geo-referencing scans to appropriate coordinate systems. Just as the number of scanner setups required for a project is a significant factor in overall field productivity for high-definition surveying, the number of target placements can also be a significant factor.
When scan targets are deployed, not only do they have to be physically put in place, but they often have to be "located" or identified by the scanner operator in the scanner's digital camera image or coarse scan and then separately scanned at a very high scan density so that their center point coordinates can be accurately "extracted" (i.e., calculated). Often scan targets are also surveyed conventionally with a reflectorless total station. Sometimes targets can be left in place, but often they are collected and removed from the site after they've been scanned. In my conversations with users, I've heard of anywhere from no targets used to as many as eight to ten targets used per setup! So the key from a productivity standpoint is to try to minimize the number of scan targets used while still ensuring the project's accuracy and geo-referencing requirements are met.
Just as a scanner's maximum field-of-view and useful range are key factors in determining the number of scanner setups, they are also key factors in determining the number of target setups needed for a project. For targets, however, some additional factors and variants come into play.
Maximum Field-of-View Per Scan
A scanner with a restricted vertical FOV can pose challenges for placement of targets. Consider again the example used for examining scanner setup factors. A user wants to capture the facade of a multi-story building or other tall structure and cannot back the scanner far enough to capture the whole vertical length of the facade in a single scan. If the scanner has an insufficient vertical FOV, then the operator might have to try to tilt up or re-orient the scan head to scan above. As soon as the scan head orientation is changed to capture the "above" geometry, then a) tilt compensation, which can otherwise reduce the number of targets needed, can no longer be used and b) this second (and third, etc.) scan will need to be separately registered to each other, preferably using targets.
But which targets? Can enough targets that were in the field-of-view of the first "non-tilted" scan still be captured with the new "tilted" or "re-oriented" scanner head? If there are enough targets, are they in locations that will provide a good network for scan registration and geo-referencing? When the scan head is re-oriented to tilt or face up, scan targets may need to be placed high up on the structure. In contrast, if the scanner has a large vertical FOV that does not require "re-orientation" or repositioning of the scan head, then the user doesn't have to worry about adding extra targets or placing targets high up on the structure.
If it's not feasible to place targets high up on the structure, then the user might try to use common features (such as window corners) of overlapping scans to try to register scans together. But this takes extra time and may not be as accurate as desired. So a good vertical FOV can provide valuable logistical benefits with regard to scan registration.
Useful Range for Scanning Targets
The key question here is "What is the farthest range at which I can scan a target to support the required project accuracy?" In general, the farther away that a target can be placed from the scanner such that a user can still extract its center with high accuracy (e.g., ± 2mm), the better. This not only directly affects the number of scanner setups, but users will be better able to place targets where they can be seen from multiple scanner setup locations, leading directly to needing fewer targets.
The limited range at which targets can be placed has historically been one of the primary field productivity challenges when using ultra-high speed, phase-based scanners. Until recently, targets often had to be placed within 10m of a phase-based scanner to achieve the desired registration accuracy. The ability of some time-of-flight scanners to capture targets with high accuracy at distances out to 50m or 100m or even farther gives users a lot of additional flexibility in the field, and experienced users are quick to point this out. It also gives users better network geometry for higher accuracy registration.
The "target capture range" of certain next generation phase scanners has been extended from 10m or so to 25m or farther, depending on the required accuracy. This extension of target capture range is expected to make phase-based scanning economically viable for more projects.
Useful Range for Scanning Other Fine Registration Features
In addition to scan targets, many people also use certain features already in the scene, such as building edges and corners, poles, steel elements, etc., to tie overlapping scans to each other. In this case, the ability to conduct very fine scanning with a small spot size and high accuracy scan data at long range is a big plus. Scanners without these capabilities often have to be moved around the site more frequently to get close enough to the scene of interest to be able to conduct such fine scans with smaller spot sizes. Users will sometimes conduct multiple scans of the exact same scene from a single scanner setup to increase scan density, which also helps extract fine features.
Small spot sizes are achieved by various means (Professional Surveyor, October 2006) but spot sizes as they exit the scanner are inherently based on the scanner's laser wavelength. So, scanners with smaller laser wavelengths can take advantage of smaller laser beams, thus maximizing the user's ability to scan targets and other fine features farther from the scanner. All of this increases field productivity.
Survey-Grade, Dual-Axis (Tilt) Sensing and Compensation
This capability in laser scanners can altogether directly reduce the number of scan targets that need to be placed, scanned, and surveyed for projects. It can also reduce or even totally eliminate office registration steps.
Essentially, the scanner takes advantage of its ability to sense and, in some cases, automatically adjust (with survey-grade accuracy) for changes in the scanner's horizontal and vertical orientation relative to true vertical. This dual-axis compensation capability is an inherent feature of total stations that most users take for granted. However, its addition to laser scanners is a relatively recent advancement. With this capability, scanner operators no longer have to place several targets in the scanner's field-of-view as a means of tying scans to each other and/or to local coordinate systems. Instead, a scanner operator can set up and level the scanner over a known point and then backsight, resection, or traverse with the scanner similar to how total stations are used. This can greatly reduce the number of scan targets needed. Registration is done automatically in the scanner control software. In its fullest implementation, dual-axis compensation can reduce the number of targets required to a minimum of one per project. This capability is also beneficial on sites where it may otherwise be difficult to place targets in a good geometric array for accurate registration.
Users of these systems have reported significant field productivity gains— on the order of 25 percent—by sharply reducing targeting requirements. They have also reported using smaller crew sizes and, in some cases, avoiding having to bring along a total station for surveying targets. Although dual-axis compensation can reduce the need for scan targets, many users still use targets for QA purposes, but they just don't use as many as they used to. From the standpoints of redundancy and overall project accuracy, the use of multiple targets is still very good practice. So some people still use the same number of targets they always have and use dual-axis compensation as an extra sensing/QA tool.
Cloud-to-Cloud Registration Software
Although not part of the scanner itself, the use of this type of software can also dramatically reduce the need to place and scan targets. It basically aligns overlapping scans with each other without the need for targets and applies best-fit algorithms to the overlapping Dual-axis compensators in scanners allow traditional survey workflows such as traversing and re-sectioning that sharply reduce the need for targets scan areas so as to minimize the spacing between the data sets in the areas of overlap.
I am familiar with some projects that have used this methodology exclusively and have thus completely avoided the use and placement of targets. However, it is more common that cloud-to-cloud registration software is used in conjunction with targets and/or with scanners with dual-axis compensation for optimizing registration and geo-referencing results. Not all vendors of laser scanning systems and software offer this type of cloud-to-cloud registration software, and, furthermore, today the capabilities of this software can vary significantly from vendor to vendor. In general, the more powerful the cloud-to-cloud registration software is, the more that users can take advantage of this workflow option in the field to minimize the number of targets needed and to optimize the accuracy of the final registration results.
Target Materials and Geometry
Vendors offer a wide range of scan target types. Specific target geometry can affect both field productivity and registration/geo-referencing accuracy. In general, larger targets (e.g., 6" diameter planar or spherical targets) are easier to identify within an image or coarse scan for the purpose of fine-scanning to extract center point coordinates. Larger targets are, however, generally more costly. Some users take advantage of inexpensive, adhesive- backed targets that can be left in place after the high-definition survey. This practice eliminates the field time needed to re-collect targets and can be helpful if the site needs to be revisited for additional survey work, as the targets can be reused without having to set them out again.
Various factors affect field productivity for high-definition surveys. Some are generic to all types of surveying, but others, such as considerations for minimizing the number of scan targets needed, are unique to scanning. Scanner capabilities— such as a full dome field-of-view, good target capture range, and dual-axis tilt compensation—plus advanced software capabilities that allow robust cloud-to-cloud registration all help to minimize the number of targets needed and maximize field productivity.
About the Author
Geoff JacobsGeoff is senior vice president, strategic marketing for Leica Geosystems, Inc.
» Back to our February 2007 Issue