PDAL Pipeline Documentation

PDAL Pipeline pdal_deli_1_1_3.py

This pipeline provides an approach to processing LiDAR data, from initial classification reset through conditional reclassification based on ground and height analysis.

General Structure and Functionality

• readers.las:

  • Initializes reading from a LAS file.
  • filename: Specifies the path to the input LAS file.

• filters.assign:

  • This filter is used to initially set all points' classifications to 0.
  • assignment directive: Classification[:]=0 applies this classification universally across all points, effectively resetting any pre-existing classifications.

• filters.smrf (Simplified Morphological Filter):

  • Used to identify ground points from non-ground points based on their geometric features.
  • Parameters: scalar, slope, threshold, window, cell are configured to tailor the filter's sensitivity and accuracy.

• filters.hag_nn (Height Above Ground Nearest Neighbor):

  • Calculates each point's height relative to the nearest ground points.
  • count: Specifies how many nearest neighbors to consider.

• Another filters.assign:

  • Uses the value parameter to conditionally reclassify points based on their calculated Height Above Ground (HeightAboveGround).
  • Points are classified as 2 if they are 1 meter or less above ground, and as 5 if they are more than 1 m above ground.

• writers.las:

  • Writes the processed and reclassified point cloud to an output LAZ file.
  • filename: Specifies the path to the output file.

• run_pipeline:

  • Takes the JSON string that defines the pipeline, converts it into a PDAL pipeline object, and executes it.

PDAL Pipeline pdal_deli_1_1_4.py

This script processes a classified point cloud (modified output from the first pipeline in .las format) by removing noise specifically classified as noise (class 7) and thinning the data to manage its density.

General Structure and Functionality

• LAS File Reader (readers.las):

  • Loads the LAS file containing the point cloud data.
  • filename: Specifies the path to the LAS file to be processed.

• Noise Removal (filters.outlier):

  • Uses a statistical approach to identify and remove outliers based on the distribution of nearest neighbor distances.
  • mean_k: Number of nearest neighbors to consider (set to 8), which influences the calculation of mean distance and standard deviation.
  • multiplier: Determines the threshold for classifying points as outliers, set to 3 times the standard deviation.

• Reclassification of Noise Points (filters.assign):

  • Reclassifies points previously identified as noise (class 7) to unclassified (class 0).
  • Uses a conditional statement to change the classification of noise points.

• Data Thinning (filters.decimation):

  • Reduces the overall number of points in the dataset to decrease processing load and improve manageability.
  • Retains every tenth point (step = 10), effectively decimating the dataset.

• LAS File Writer (writers.las):

  • Saves the processed point cloud to a new LAZ file.
  • filename: Sets the destination for the output file.

• Execution Function (run_pipeline):

  • A function that takes the defined JSON pipeline, converts it into a PDAL pipeline object, and executes the processing steps.

PDAL Documentation