Aller au contenu principal

RGB Augmentation API Reference

The RGB Augmentation API provides tools for integrating IGN orthophoto data with LiDAR point clouds to create color-enhanced datasets.

Core Classes​

RGBProcessor​

Main class for RGB augmentation operations.

from ign_lidar import RGBProcessor

processor = RGBProcessor(
    interpolation_method='bilinear',
    quality_threshold=0.8,
    enable_caching=True
)

Methods​

augment_point_cloud(points, orthophoto_path)​

Adds RGB values to point cloud from orthophoto.

Parameters:

  • points (numpy.ndarray): Point coordinates (N×3)
  • orthophoto_path (str): Path to orthophoto file

Returns:

  • numpy.ndarray: RGB values (N×3) as uint8

batch_augmentation(tile_list, ortho_dir)​

Process multiple tiles with RGB augmentation.

Parameters:

  • tile_list (list): List of tile paths
  • ortho_dir (str): Directory containing orthophotos

Returns:

  • dict: Results dictionary with augmented data

Configuration Options​

Interpolation Methods​

  • nearest: Fastest, pixel-exact colors
  • bilinear: Smooth color transitions
  • bicubic: Highest quality interpolation

Quality Control​

processor = RGBProcessor(
    quality_threshold=0.9,  # Color accuracy threshold
    validate_coordinates=True,  # Check point-photo alignment
    handle_missing_data=True   # Fill gaps gracefully
)

Error Handling​

try:
    rgb_data = processor.augment_point_cloud(points, orthophoto)
except OrthophotoNotFoundError:
    print("Orthophoto file not accessible")
except CoordinateMismatchError:
    print("Point cloud and orthophoto coordinates don't align")
except InsufficientOverlapError:
    print("Not enough overlap between data sources")

Performance Optimization​

GPU Acceleration​

processor = RGBProcessor(
    use_gpu=True,
    gpu_batch_size=50000,
    enable_gpu_caching=True
)

Memory Management​

# Process large datasets efficiently
def process_large_dataset(points, orthophoto):
    chunk_size = 100000
    rgb_results = []

    for i in range(0, len(points), chunk_size):
        chunk = points[i:i+chunk_size]
        rgb_chunk = processor.augment_point_cloud(chunk, orthophoto)
        rgb_results.append(rgb_chunk)

    return np.concatenate(rgb_results)

Examples​

Basic RGB Augmentation​

import numpy as np
from ign_lidar import RGBProcessor

# Initialize processor
processor = RGBProcessor()

# Load point cloud
points = np.load('building_points.npy')

# Add RGB data
rgb_colors = processor.augment_point_cloud(
    points,
    'orthophoto.tif'
)

# Save enhanced point cloud
enhanced_points = np.column_stack([points, rgb_colors])

Batch Processing with Quality Control​

processor = RGBProcessor(
    quality_threshold=0.85,
    validate_coordinates=True
)

results = processor.batch_augmentation(
    tile_list=['tile1.las', 'tile2.las'],
    ortho_dir='/path/to/orthophotos/'
)