What is TerraMatch?
Terrasolid’s TerraMatch module works in conjunction with TerraScan to allow users to perform boresight calibrations and apply geometric corrections to point cloud data. Originally developed to handle LIDAR data sets from fixed-wing and helicopter platforms, TerraMatch has evolved into a set of powerful general-purpose tools for working with any point cloud data where kinematic – in motion – lidar sensors need to have systematic and dynamic errors removed from the point cloud. TerraMatch is sensor-agnostic and can be used with any point cloud data that includes the post-processed trajectory information from the sensor.
Callibration vs. Dynamic Error
Before discussing the TerraMatch tools and workflow, it is important to distinguish between “calibration” and “dynamic error” when talking about LIDAR systems.
Often when there are shifts in the point cloud data after post-processing is completed, users will say “the calibration is off” or “we need to re-do the calibration”, but this is usually not the case at all. All LIDAR systems undergo a rigorous system calibration at the factory. This factory calibration typically solves for the intrinsic relation between elements of the sensor platform such as position of the GNSS antenna, the alignment of the laser with respect to the Inertial Measurement Unit (IMU) body frame, the internal offsets or lever arms between the main subsystems, camera alignment to this same IMU body frame, intrinsic camera parameters such as principal point, focal length and so on. These are static parameters for the system that do not change significantly over time.
With our True View drone LIDAR/imagery sensors, we solve for these parameters in our factory calibration processes. We have learned how to do this to a high degree of accuracy and these parameters (due to our rigid platform designs) hold exceptionally well over time. These calibration parameters are maintained in the database for each sensor and automatically accessed via the Reckon web portal during post-processing so that the latest, most accurate calibration file is always used by True View EVO software. The common characteristic of calibration parameters is that they are fixed, measurable and can be systematically corrected for during post-processing.
In contrast to the calibration of the LIDAR sensor – the careful measurement of the static parameters that define the sensors rigid geometry and timing errors – dynamic errors are present in, and can vary from, project to project.
Sources of dynamic error can include GNSS position errors, IMU angular errors in roll, heading and pitch (with heading being the most problematic), vertical bias and, of course, blunders and human error.
Dynamic errors, unlike static calibration parameters, have the unfortunate tendency that they can be time dependent. This is why, for example, flight lines from different lifts/flights on different days may exhibit an offset or shift between them. The sensor calibration did not change – assuming a rigid sensor and no crashes in between lifts! – but there is often a residual offset between the two data sets due to the different dynamic errors present during the data collection. Minimizing these offsets by performing a strip adjustment is where TerraMatch is used.
Strip Adjustment with TerraMatch
TerraMatch uses the traditional approach to strip adjustment of identifying common tie lines and tie points in each lidar flight line and fitting these tie lines and tie points to minimize the dynamic errors between the lines and between separate lifts/flights. This approach can solve for various errors including shifts in heading, roll and pitch (hrp), vertical biases between lines (dz), vertical biases between groups (lifts/flights) (dZ), planimetric shifts (dxdy) and time-dependent variations (or fluctuations) in all the above. Our experience has shown that given the rigid design and rigorous factory calibration, most projects flown with our True View sensors do not require any major strip adjustment beyond a basic angular adjustment and an elevation shift, a “hrpdz” matching. The offsets are usually most noticeable between lifts/flights, especially if there is significant time between collections. In these cases, we recommend using TerraMatch for strip alignment.
The basic TerraMatch workflow is as follows:
- Import and set-up trajectories and point cloud data. Verify coordinate system, height reference, and timeframe/format.
- Classify ground by flight line. Classify building roofs by flight line if desired/available.
- Review data using Compute Distance (Ground to Line Average Z) and Measure Match.
- Search for tie lines using Surface Lines (Slope Direction). Review results and tie line distribution.
- Solve for a hrpdZ (heading, roll, pitch and dZ) correction by Group (Lift/Flight).
- Apply By Group corrections.
- Solve for a hrpdz (heading, roll, pitch and dZ) correction by individual Flight Line.
- Apply By Flight Line corrections.
- Review results.
- If necessary, use Find Fluctuations to apply a time-dependent drift corrections to each flight line.
TerraMatch includes a set of reporting tools for quantifying the fit or “match” between lines and between lifts/flights. These allow the user to monitor progress and quantify the amount by which they are improving the overall fit of the data.
The primary measure of fit in TerraMatch is the average mismatch between the tie lines and tie points from different lines reported as a 3d, planimetric (XY) or vertical (Z) mismatch. For example, here are the reported mismatches for True View 515 data in sloped, forested terrain before and after a hrpdZ correction:
Before (Units Meters)
After (Units Meters)
In this case there was an overall improvement of ~25% in the matching between flight lines by removing the residual dynamic error even though the original, unadjusted data was already within system specifications with a 2.1 cm average Z and 4.4 cm average 3d matching between lines.
Qualitative checks can always be done by visually inspecting the overlapping flight lines, ideally in areas of sloped features that highlight the dynamic error. For example, here are cross-sections of the True View 515 data taken in the area between two different lifts on two different days, positioned along the edge of a highway before and after a TerraMatch correction:
Before TerraMatch Correction
After Terramatch Correction The residual shift between lines has been minimized and the lines now fit within the noise (precision) envelope of the True View 515.