ferrum_cloud

Pure Rust implementation of Point Cloud Library (PCL).

This library provides efficient, safe, and ergonomic APIs for point cloud processing, leveraging Rust's ownership system, zero-cost abstractions, and parallel processing capabilities.

Features

• Memory Safety: Leverages Rust's ownership system for safe memory management
• Zero-Copy Operations: Efficient processing through views and references
• Parallel Processing: Built-in support for parallel operations using Rayon
• Generic Point Types: Flexible point type system supporting various point formats
• Comprehensive I/O: Support for multiple point cloud file formats (PCD, PLY, LAS)

Quick Start

use ferrum_cloud::prelude::*;

fn main() -> Result<()> {
    // Load a point cloud
    let cloud = io::load_pcd("examples/scene.pcd")?;

    // Process the cloud
    let processed = cloud
        .voxel_downsample(0.05)
        .remove_outliers(50, 1.0)?;

    // Save the result
    io::save_ply(&processed, "examples/processed.ply")?;

    Ok(())
}

Examples

Creating Point Clouds

use ferrum_cloud::prelude::*;

fn main() -> Result<()> {
    // Create a simple XYZ point cloud
    let points = vec![
        PointXYZ::new(0.0, 0.0, 0.0),
        PointXYZ::new(1.0, 0.0, 0.0),
        PointXYZ::new(0.0, 1.0, 0.0),
        PointXYZ::new(0.0, 0.0, 1.0),
    ];
    let cloud = PointCloud::from_points(points);

    // Create a point cloud with RGB colors
    let colored_points = vec![
        PointXYZRGB::new(0.0, 0.0, 0.0, 255, 0, 0),   // Red
        PointXYZRGB::new(1.0, 0.0, 0.0, 0, 255, 0),   // Green
        PointXYZRGB::new(0.0, 1.0, 0.0, 0, 0, 255),   // Blue
    ];
    let colored_cloud = PointCloud::from_points(colored_points);

    Ok(())
}

Filtering Point Clouds

use ferrum_cloud::prelude::*;

fn main() -> Result<()> {
    let cloud = io::load_pcd("examples/scene.pcd")?;

    // Voxel downsampling - reduce density while preserving structure
    let downsampled = cloud.voxel_downsample(0.05);

    // Statistical outlier removal - remove noisy points
    let filtered = downsampled.remove_outliers(50, 1.0)?;

    // Radius outlier removal - remove isolated points
    let clean = filtered.remove_radius_outliers(0.5, 3);

    // Pass-through filter - crop along an axis
    let cropped = clean.pass_through(Axis::Z, 0.0, 2.0);

    Ok(())
}

I/O Operations

use ferrum_cloud::prelude::*;

fn main() -> Result<()> {
    // Load from different formats
    let pcd_cloud = io::load_pcd("input.pcd")?;
    let ply_cloud = io::load_ply("input.ply")?;
    let las_cloud = io::load_las("input.las")?;

    // Save to different formats
    io::save_pcd(&pcd_cloud, "output.pcd")?;
    io::save_ply(&ply_cloud, "output.ply")?;
    io::save_las(&las_cloud, "output.las")?;

    Ok(())
}

Working with Point Types

use ferrum_cloud::prelude::*;

fn main() -> Result<()> {
    // Access point coordinates
    let point = PointXYZ::new(1.0, 2.0, 3.0);
    let x = point.x();
    let y = point.y();
    let z = point.z();
    let pos = point.position(); // [f32; 3]

    // Calculate distances
    let p1 = PointXYZ::new(0.0, 0.0, 0.0);
    let p2 = PointXYZ::new(3.0, 4.0, 0.0);
    let dist = p1.distance_to(&p2); // 5.0

    // Work with colored points
    let rgb_point = PointXYZRGB::new(1.0, 2.0, 3.0, 255, 128, 64);
    let packed_rgb = rgb_point.rgb(); // 0xFF8040
    let normalized = rgb_point.rgb_normalized(); // [1.0, 0.502, 0.251]

    Ok(())
}

Point Cloud Information

use ferrum_cloud::prelude::*;

fn main() -> Result<()> {
    let cloud = io::load_pcd("examples/scene.pcd")?;

    // Get cloud information
    let num_points = cloud.len();
    let is_empty = cloud.is_empty();

    // Iterate over points
    for point in cloud.iter() {
        let pos = point.position();
        // Process each point...
    }

    // Parallel iteration for performance
    use rayon::prelude::*;
    cloud.par_iter().for_each(|point| {
        // Parallel processing...
    });

    Ok(())
}

Supported Point Types

• PointXYZ: Basic 3D point with x, y, z coordinates
• PointXYZRGB: 3D point with RGB color (r, g, b as u8)
• PointXYZRGBNormal: 3D point with RGB color and normal vector

Supported File Formats

• PCD: Point Cloud Data (PCL format)
• PLY: Polygon File Format (Stanford format)
• LAS: LiDAR point cloud format

License

MIT