Add pointcloud to mesh pipeline

pointcloud_to_mesh/README.md

@@ -0,0 +1,142 @@
+# Pointcloud to Mesh Pipeline
+
+A modular pipeline for converting LiDAR and point cloud scans into optimized 3D geometry.
+
+This module is designed for experimental workflows involving scan cleanup, voxel processing, clustering, and mesh preparation for fabrication, visualization, robotics, and digital design workflows.
+
+---
+
+# Features
+
+* Point cloud cleanup and preprocessing
+* Voxel downsampling
+* Cluster detection
+* Bounding box generation
+* Mesh preparation pipeline
+* Visualization previews with Open3D
+* Experimental scan segmentation workflow
+
+---
+
+# Project Structure
+
+```text
+pointcloud_to_mesh/
+├── data/
+│   └── input point clouds
+├── output/
+│   └── generated meshes and exports
+├── mesh.py
+├── voxel_test.py
+├── requirements.txt
+└── README.md
+```
+
+---
+
+# Requirements
+
+Python 3.10+
+
+Install dependencies:
+
+```bash
+pip install -r requirements.txt
+```
+
+---
+
+# Dependencies
+
+Main libraries used:
+
+* Open3D
+* NumPy
+* SciPy
+* Trimesh
+
+---
+
+# Usage
+
+Run the mesh pipeline:
+
+```bash
+python mesh.py
+```
+
+Run voxel testing:
+
+```bash
+python voxel_test.py
+```
+
+---
+
+# Pipeline Overview
+
+## 1. Point Cloud Cleanup
+
+The input point cloud is cleaned and filtered to remove noise and reduce unnecessary geometry.
+
+## 2. Voxel Downsampling
+
+Voxelization reduces the density of the point cloud while preserving overall geometry.
+
+## 3. Cluster Detection
+
+Clusters are detected from spatially connected point regions.
+
+## 4. Bounding Box Generation
+
+Oriented bounding boxes are generated for segmented clusters.
+
+## 5. Mesh Preparation
+
+The processed geometry can be used for:
+
+* mesh reconstruction
+* fabrication workflows
+* robotic processing
+* visualization
+* spatial analysis
+
+---
+
+# Example Applications
+
+* LiDAR scanning workflows
+* Robotic fabrication
+* Architectural scanning
+* Spatial analysis
+* Experimental digital fabrication
+* Interactive installations
+* 3D reconstruction research
+
+---
+
+# Notes
+
+This project is experimental and currently under active development.
+
+Some datasets may require preprocessing depending on scan quality and density.
+
+---
+
+# Future Improvements
+
+Planned additions:
+
+* Automatic mesh reconstruction
+* Surface smoothing
+* Better segmentation
+* GPU acceleration
+* Real-time visualization
+* Export tools for Rhino / Grasshopper
+* Better scan optimization
+
+---
+
+# Author
+
+Developed as part of an experimental point cloud and spatial geometry workflow.

pointcloud_to_mesh/mesh.py

@@ -0,0 +1,114 @@
+from pathlib import Path
+import open3d as o3d
+import numpy as np
+
+# PATHS
+base_dir = Path(__file__).parent
+
+input_path = base_dir / "data" / "last.ply"
+output_path = base_dir / "output" / "last_mesh.obj"
+
+output_path.parent.mkdir(exist_ok=True)
+
+# LOAD
+print("Loading point cloud...")
+pcd = o3d.io.read_point_cloud(str(input_path))
+
+# SCALE
+scale_factor = 1000
+pcd.scale(scale_factor, center=(0, 0, 0))
+
+print("Point count:", len(pcd.points))
+
+if len(pcd.points) == 0:
+    raise ValueError("Point cloud is empty")
+
+# SCALE INFO
+bbox = pcd.get_axis_aligned_bounding_box()
+scale = np.linalg.norm(bbox.get_extent())
+
+# SETTINGS
+voxel_size = scale / 350
+normal_radius = scale / 60
+
+# DOWNSAMPLE
+print("Downsampling...")
+pcd = pcd.voxel_down_sample(voxel_size)
+
+# OUTLIER REMOVAL
+print("Removing outliers...")
+pcd, _ = pcd.remove_statistical_outlier(
+    nb_neighbors=30,
+    std_ratio=1.5
+)
+
+# NORMALS
+print("Estimating normals...")
+pcd.estimate_normals(
+    search_param=o3d.geometry.KDTreeSearchParamHybrid(
+        radius=normal_radius,
+        max_nn=50
+    )
+)
+
+try:
+    pcd.orient_normals_consistent_tangent_plane(50)
+except Exception as e:
+    print("Normal orientation skipped:", e)
+
+# POISSON
+print("Generating mesh...")
+mesh, densities = o3d.geometry.TriangleMesh.create_from_point_cloud_poisson(
+    pcd,
+    depth=9
+)
+
+# DENSITY CLEANUP
+print("Cleaning mesh...")
+
+densities = np.asarray(densities)
+density_threshold = np.quantile(densities, 0.05)
+
+vertices_to_remove = densities < density_threshold
+mesh.remove_vertices_by_mask(vertices_to_remove)
+
+# REMOVE SMALL FLOATING COMPONENTS
+triangle_clusters, cluster_n_triangles, cluster_area = (
+    mesh.cluster_connected_triangles()
+)
+
+triangle_clusters = np.asarray(triangle_clusters)
+cluster_n_triangles = np.asarray(cluster_n_triangles)
+
+min_triangles = 5000
+triangles_to_remove = cluster_n_triangles[triangle_clusters] < min_triangles
+
+mesh.remove_triangles_by_mask(triangles_to_remove)
+mesh.remove_unreferenced_vertices()
+
+# FINAL CLEANUP
+mesh.remove_degenerate_triangles()
+mesh.remove_duplicated_triangles()
+mesh.remove_duplicated_vertices()
+mesh.remove_non_manifold_edges()
+mesh.remove_unreferenced_vertices()
+
+# SMOOTH EDGES / SURFACE
+print("Smoothing mesh...")
+mesh = mesh.filter_smooth_simple(number_of_iterations=2)
+
+# CLEAN AGAIN AFTER SMOOTH
+mesh.remove_degenerate_triangles()
+mesh.remove_duplicated_triangles()
+mesh.remove_duplicated_vertices()
+mesh.remove_non_manifold_edges()
+mesh.remove_unreferenced_vertices()
+
+mesh.compute_vertex_normals()
+
+# SAVE
+print("Saving OBJ...")
+o3d.io.write_triangle_mesh(str(output_path), mesh)
+
+print("DONE")
+print(output_path)