Efficient point cloud extraction from waveform data

BayesWavEx 1.0
Goals:

Extract 3D points from raw, full waveform LiDAR files. Get more points and more accurate results even at high altitude and in difficult cases with low vegetation. High throughput achieved using efficient algorithms and an optimized implementation.

Input:

RAW waveform file (LAS 1.3 FWF)

Output:

– LAS/LAZ 1.2 point cloud [extra attributes: uncertainty, width…]
– [estimated trajectory, statistics]

Supported platforms:

Windows, MacOS – 64-bit – Request a demo

Documentation [pdf, 104kB, 9/6/16]

Advantages:
  • Ground extraction robust to peak contamination and overlaps
  • Underground false alarm suppression using pulse shape
  • Automatic radiometric and impulse response calibration
  • Fast, accurate waveform decomposition
  • Ultra-fast extraction options for quick preview
  • Physically meaningful target attribute extraction
  • Range / attribute uncertainty attribute export options
  • Basic outlier filtering (high/low point elimination)
  • Intensity correction options
  • No trajectory file needed
Full waveform LiDAR data processing, raw data, peak extraction, pulse detection, last return, range estimation, waveform decomposition, Gaussian decomposition, signal processing

Full-waveform data processing framework:
Raw binary files are read, decoded and processed to extract a maximum number of peaks in a robust way, despite contamination by noise, sensor artifacts and low vegetation. Outlier filtering is run on each scanline to reject low and high points. The trajectory is reconstructed using available georeferencing information and used for optional range-based intensity correction. Range uncertainty is computed using extracted peak parameters.
Accurate gridded products can then be generated with the upcoming BayesAccuGrid module taking full advantage of the computed pointwise uncertainty, for an optimal combination of swaths.

Scanline-based ground peak extraction: This graph shows the raw waveform data part of a scanline, and the extracted ground points. Notice the robustness to the ringing artifact on the left. The ground returns predicted by the probabilistic model are shown in brown, and the rejected outliers in green.

Scanline-based ground peak extraction:
This graph shows the raw waveform data part of a scanline, and the extracted ground points. Notice the robustness to the ringing artifact on the left. The ground returns predicted by the probabilistic model are shown in brown, and the rejected outliers in green.
The ground prediction feature will be available in Version 1.1; the current version uses a simpler outlier rejection model.

Physical parameters such as range and target thickness are estimated using a waveform decomposition that takes into account the system impulse response. This way a low false alarm rate and optimal location accuracy can be achieved.

Physical parameters such as range and target thickness are estimated using a waveform decomposition that takes into account the system impulse response. This way a low false alarm rate and optimal location accuracy can be achieved.

Your benefits

Accuracy improvement

Bayesian inference is not only about producing fuzzy quantities! It actually ensures that results have maximum accuracy, compared to ad-hoc or classical approaches. Proper modeling ensures reduced errors and is effective in waveform processing. High detection rates can be achieved, allowing you to fly higher and cut collection costs!

Optimal data fusion, or swath combination, helps produce highest possible accuracy elevation models given the collected data.
full waveform LiDAR processing, last return, echo extraction, pulse detection, Gaussian peak, laser ranging

Automation

Manually tuning essential processing parameters is no longer necessary. Topography smoothness, outlier frequency, that’s the algorithm’s problem, not yours!

Deciding at what scan angle to cut off data is over! Uncertainty propagation ensures a proper weighting is applied to each data point, avoiding throwing away precious information and significant portions of the input point clouds.
automatic parameter estimation, unsupervised

Uncertainty management

Managing uncertainty properly allows not only to propagate errors, but to preserve information by using error structure (e.g. covariances). Understanding how uncertainty affects elevation is essential for further shape analysis and change detection. This does not come with the data, but we help you compute it!
error propagation, uncertainty management


Example of attributes exported by WavEx: range uncertainty (top), target thickness (middle) and corrected intensity (bottom). The reconstructed trajectory points are shown above the three layers. NERC-ARSF data, see this blog page for details.

Example of attributes exported by WavEx: range uncertainty (top), target thickness (middle) and corrected intensity (bottom). The reconstructed trajectory points are shown in white above the three layers. NERC-ARSF data, see this blog page for details.

Impulse response of the Leica ALS-50 LiDAR waveform digitizer related to the NERC-ARSF data in the example above. The software estimates it from random samples taken from the waveform file in a robust way without user interaction.

Impulse response of the Leica ALS-50 LiDAR waveform digitizer related to the NERC-ARSF data in the example above. The software estimates it from random samples taken from the waveform file, in a robust way, no user interaction required.