3D-Mesh-Measurement-System

A Python-based 3D geometry system that reads raw .obj mesh files, extracts vertex data, applies Principal Component Analysis (PCA) to determine object orientation, computes a tight Oriented Bounding Box (OBB), and visually validates dimensions and volume using Open3D.

---

Table of Contents

1. Overview
2. Features
3. Folder Structure
4. How to Run Locally
5. System Flow & Design Decisions
6. Geometry Processing Pipeline
7. Mermaid Flow Diagram
8. Visualization & Validation Strategy
9. Challenges & Trade-Offs
10. Performance & System Notes

---

Overview

This project implements an end-to-end 3D object measurement pipeline designed to compute accurate dimensions of raw 3D mesh objects.

The system:

• Reads .obj mesh files
• Converts mesh geometry into a vertex-based point cloud
• Applies PCA to identify the object’s natural orientation
• Computes a tight Oriented Bounding Box (OBB)
• Outputs length, width, height, and volume
• Visually validates correctness through interactive 3D rendering

The approach reflects real-world workflows used in 3D scanning, CAD inspection, robotics preprocessing, and packaging optimization systems.

---

Features

• Raw .obj mesh ingestion
• Vertex-based point cloud processing
• PCA-based orientation estimation
• Tight OBB dimension and volume computation
• Interactive 3D visualization using Open3D
• Deterministic, CPU-only geometry pipeline

---

Folder Structure

part1/
│
├── data/
│   └── .obj files
│
├── main.py
└── README.md

---

How to Run Locally

1. Create Virtual Environment

python3 -m venv venv
source venv/bin/activate

2. Install Dependencies

pip install numpy open3d

3. Run the Program

python3 main.py

---

System Flow & Design Decisions

The system follows a linear geometry-processing flow, where each stage feeds directly into the next.

---

Mermaid Flow Diagram

flowchart TD

A[OBJ Mesh File] --> B[Open3D Mesh Loader]
B --> C[Vertex Extraction]
C --> D[Point Cloud Centering]

D --> E[PCA Computation]
E --> F[Eigenvalues & Eigenvectors]

F --> G[Rotate Points into PCA Frame]
G --> H[Min-Max Extent Computation]

H --> I[OBB Dimensions & Volume]
I --> J[Open3D Visualization]

J --> K[Interactive Validation]

Loading

---

Geometry Processing Pipeline

1. Mesh Ingestion

• .obj file is loaded using Open3D
• Mesh vertices are extracted for geometric processing

2. Point Cloud Centering

• Centroid of all vertices is computed
• Point cloud is centered to ensure PCA stability

3. PCA Orientation Estimation

• Covariance matrix is computed from centered points
• Eigen decomposition yields principal axes
• Eigenvectors define object orientation

4. Oriented Bounding Box Computation

• Points are rotated into PCA-aligned frame
• Min–max extents are computed
• Dimensions (L × W × H) and volume are calculated

---

Visualization & Validation Strategy

Visualization is treated as proof of correctness, not decoration.

• Original mesh rendered in gray

• Oriented Bounding Box rendered in red

• Interactive camera rotation validates:

  • Correct PCA orientation
  • Tight bounding box fit
  • Absence of axis-aligned bias

This step acts as a visual unit test for the geometry pipeline.

---

Challenges & Trade-Offs

1. Orientation Ambiguity

• Raw meshes may be arbitrarily rotated
• PCA ensures deterministic orientation handling

2. Bounding Box vs True Volume

• System computes bounding box volume, not exact mesh volume
• Chosen to match industrial packaging and inspection use-cases

3. Visualization Backend

• Open3D relies on OpenGL/X11
• Wayland sessions may require Xorg fallback

---

Performance & System Notes

• CPU-only computation (no CUDA/GPU dependency)
• PCA operates on a 3×3 covariance matrix
• Suitable for low-resource systems
• Scales linearly with vertex count

---