Strip Align - Method and Context

Method

Strip alignment, also known as strip align, strip matching, strip adjustment, line matching, or point cloud registration, is a common practice in lidar data processing.  Strip alignment techniques are found wherever dynamic remote sensing measurement systems are being used.  In LP360, the term 'strip align', and the tool Strip Align, are used interchangeably to refer to this method of data correction.  The goal of strip alignment is to have a robust method to remove inconsistencies in the point cloud caused by dynamic errors in the sensor trajectory and minimize any residual boresight errors.  Such errors cause objects scanned at different times or in different passes of the same object to appear shifted from each other in the point cloud.  The peak of a roof being shifted from flight line to flight line is a common visual example of such errors in airborne datasets.  

Errors in the boresight angles of any lidar system have considerable impacts on the point cloud and will cause large displacements between lines/scans if not correctly established.  In addition, the positional error introduced by boresight errors is proportional to the measurement range.  It is important to minimize these angular errors as much as possible.  In a rigid system, boresight angle errors are systematic (unchanging) and can be minimized by applying a fixed set of angular corrections in heading, roll, and pitch across the entire dataset.  However, errors in the point cloud can also be induced by dynamic position and orientation errors in the trajectories.  Trajectory errors can be time-varying and therefore an estimate of a constant correction, similar to a boresight angle correction, would not be effective to remove inconsistencies in the point cloud caused by trajectory errors.  To overcome this, strip alignment methods include time-dependent corrections to the trajectory that reduce or eliminate this error dynamically along the trajectory.  In LP360's Strip Align, these dynamic trajectory corrections are based on an ICP (Iterative Closest Point) approach to matching.

The strip alignment method developed by GeoCue is based on ICP and the goal is to estimate and minimize the errors in the trajectory.  The ICP method is rigorous since real points are considered, and the model also considers the geolocation of points using the georeferenced point cloud.  The uncertainty associated with each point is also computed from uncertainties of the raw data: this enables a reduction of the impact of uncertainties related to the system on the estimated trajectory correction.

The LP360 Strip Align correction method is implemented in such a way that it is possible to define a reference surface or reference flight(s) that are considered 'known good' in advance, although this is not required.  If defined, the reference surface/flights will be used by Strip Align in the analysis but not adjusted.  The associated trajectories for the fixed reference lines will also not be adjusted.  Effectively this allows the tool to solve for a set of corrections for one group of flights and adjust them to the 'known good' reference layer. 

Strip Align is fully automatic and requires a limited number of inputs by the user, primarily the LAS layers to use for determining the correction parameters and any fixed LAS layers to use as a 'known good' reference surface. Note that the layers to be corrected must have an associated trajectory in the same LP360 project; layers used only as fixed reference layers do not need an associated trajectory layer.  When solving for a set of correction parameters and refining the trajectory, Strip Align only considers tie points in overlapping/side-lapping areas covered by 2 or more strips/passes, so overlapping passes of common objects is required for Strip Align to work. From a multi-strip survey dataset, the Strip Align tool:

• Extracts the overlapping areas.
• Preprocesses data and selects a set of Tie Points for calculation of a correction set and adjustment of the trajectories.
• Adjusts, by an advanced Least Square method, a time-dependent trajectory.
• Regenerates (geocodes) a new point cloud using the correction set calculated.

Optimal Context for Strip Alignment

In general, airborne laser systems collect data strip-wise with some overlap for full coverage of
a survey area.  Even if the lidar system is well-calibrated (for boresight, lever-arms, timing) and free of systematic errors from integration parameters, there may be time-varying navigation errors generated by the GNSS and/or IMU uncertainties and weak estimation of the trajectory.  Indeed, due to the limited accuracy of navigation data, as given by the GNSS and IMU, data gaps may be found within the overlaps between strips.  In addition, trajectory measurements are affected by external influences (number of satellites, multipath, loss of carrier phase…), thus their accuracy may dynamically change during a flight.  This causes time-dependent navigation (position and orientation) errors in the lidar point cloud.

To overcome this problem, strip alignment can be used to improve the data consistency within overlapping areas. However, to use Strip Align effectively, the sensor should already have a good initial orientation set-up with a good calibration applied to the system.  Strip Align works well to remove residual boresight errors and correct dynamic trajectory errors.  It is not intended to be used to perform a rigorous sensor calibration.  Proper boresight angles, lever arms, and potential latency should be known and applied in the sensor calibration before running Strip Align.  Remember; calibrate, collect, align.

To summarize, if after geocoding data with proper calibration parameters applied (boresight angles, levers arms, and latency) there are still undesirable inconsistencies in the point cloud, Strip Align should be used to minimize these residual errors and improve the consistency and fit of the point cloud.

See Strip Calibration.

See How to use Strip Adjustment.

Or, for some recommendations on using Strip Adjustment for Processing Multiple flights.