Extract Curves and Centroids (QGIS Processing Script)

Note

This standalone script is no longer actively developed. Its functionality is now maintained as part of L-RAT, a broader QGIS plugin for linear referencing and analysis.

This QGIS Processing script extracts curve segments from a densified line layer, calculates their radius of curvature, optionally groups nearby curve centers, and produces up to two output layers:

Curves and centroids:

• Curves layer
  Each feature represents a curve segment with attributes:

  • ID_Curva: unique ID for the curve segment
  • ID_Centroide: ID of the cluster centroid the curve belongs to (or -1 if not applicable)
  • Radio: curvature radius (m)
  • Longitud: segment length (m)

• Centroids layer (optional)
  Each feature represents a cluster of nearby curve centers, with attributes:

  • ID_Centroide: unique cluster ID
  • Radio_medio: mean radius of all curves in the cluster (m)
  • Conteo: number of curves in the cluster

---

⚙️ Parameters

• Input line layer: Vector line layer to be analyzed.
• Densification interval (m): Distance between added vertices when densifying lines (default: 15.0).
• Minimum radius (m): Exclude curves tighter than this threshold (default: 2.0).
• Maximum radius (m): Exclude curves flatter than this threshold (default: 50.0).
• Minimum distance between vertices (m): Skip triplets where vertices are too close (default: 0.5).
• Generate centroid layer (boolean): If enabled, produces a second point layer with curve centers (default: False).
• Cluster distance (m): Maximum distance for grouping curve centers into the same cluster (default: 10.0).

Script window:

---

📊 Methodology

The script analyzes every triplet of consecutive vertices (p1, p2, p3) in the densified line layer:

1. Compute edge lengths:

   • a = dist(p1, p2)
   • b = dist(p2, p3)
   • c = dist(p3, p1)

2. Compute semi-perimeter:

   • s = (a + b + c) / 2

3. Compute area using Heron’s formula:

   • area = sqrt(s(s−a)(s−b)(s−c))

4. Compute circumcircle radius:

   • R = (a·b·c) / (4·area)

   ⚠️ If points are nearly collinear (area ≈ 0), no radius is calculated.

If enabled, curve centers are also computed and then clustered based on proximity.

---

🚀 Installation

1. Download the script file Radios_curva.py.
2. In QGIS:
   • Open Processing Toolbox → Scripts → Add Script to Toolbox.
   • Load the script into your QGIS environment.

📸 Example of the installation step:

3. The algorithm will appear at your Processing Toolbox under the group: Script → Analisis de linea → Extraer curvas y centroides.

---

✅ Typical Use Cases

• Road safety studies: detecting tight bends.
• Geometric road design checks.
• Mapping and categorizing curvature for line-based infrastructures.

---

❓ Frequently Asked Questions (FAQ)

Q: Why are some curves missing in the output?

• They might not fall within the specified min/max radius range.
• Triplets of points too close together may be skipped due to the minimum distance filter.
• Collinear points do not produce a valid radius.

Q: Why do I get an empty centroid layer?

• Ensure the "Generate centroid layer" option is enabled.
• Curves must have valid centers to form clusters.

Q: What CRS should I use?

• Use a projected coordinate system (e.g., UTM) where distances are measured in meters. Using geographic CRS (lat/long) will give incorrect results.

Q: Can I adjust how many vertices are generated?

• Yes, by changing the densification interval. Smaller values create more vertices and higher precision, but increase computation time.

---

⚠️ Limitations

• The algorithm assumes reasonably clean line geometries; noisy data may produce many false curves.
• Results depend heavily on the chosen densification interval. Too coarse → missed curves; too fine → excessive computation.
• The clustering method is purely distance-based; it does not account for curve orientation or road context.
• Nearly collinear triplets of points cannot generate valid radii (they are skipped).
• Designed for projected CRS in meters — using geographic CRS (lat/long) could give meaningless distances.

---

📄 License

This project is released under the GNU General Public License v3.0. Feel free to use, modify, and share.

---

👤 Author

Javi H. Piris

• Email: [email protected]
• GitHub: @Javisionario