Smooth Adjustment

This repository provides the official C++ implementation of our paper:

"Structured Analytic Mappings for Point Set Registration"

It consists of three main components:

• Analytic-ICP DLL – Implements the core nth-order analytic ICP algorithm
• SmoothAdjustment DLL – Provides post-registration tools such as surface smoothing, visualization, and measurements
• Test Application – An executable frontend for running registration experiments and visualizing results

---

Overview

The SmoothAdjustment module is the primary framework in this project. It builds upon the Analytic-ICP library to support advanced operations on registered surfaces, such as:

• Analytic warping of surfaces
• Measurement and evaluation tools
• Visualization (2D/3D)

While currently relying on Analytic-ICP, the design is modular and can interface with other registration backends.

---

Prerequisites

To build and run this project, the following are required:

• Operating System: Windows
• Compiler/IDE: Visual Studio 2013
• Dependencies:
  • Eigen
  • Boost (for KD-tree support)
  • OpenCV (for image display in the Test app)

⚠️ Note: This project currently supports only Visual Studio 2013. Later versions may require project file adjustments.

---

Implementation Notes

• The ICP module is based on Prof. Andreas Geiger’s point-to-point ICP implementation.
• We have extended the original code by:
  • Integrating Eigen for all matrix and vector operations
  • Replacing the KD-tree implementation with Boost KD-tree
• These improvements enable the handling of large-scale point clouds in both 2D and 3D.
• The SmoothAdjustment library uses OpenCV for interactive image-based diagnostics and rendering.
• The Test application serves as a lightweight frontend to configure and run experiments.

---

Datasets

The following datasets are included in the package:

• SHREC’19 benchmark dataset
• 3DScanRep real-world scans
• Synthetic examples (2D & 3D)

All datasets used in the experiments are bundled with the release. No external downloads are required.

---

Quick Start

1. Download or clone this repository.
2. Navigate to the Release/ folder.
3. Place the required OpenCV DLLs (e.g., opencv_world300.dll) into the Release/ directory.
4. Edit the experiment.ini configuration file. Specify:
   • Path to input point sets
   • Registration parameters
   • Output path
5. Run Test.exe.

A working example is provided. After setup, simply double-click Test.exe to launch the default experiment.

---

File Structure

SmoothAdjustment/
├── AnalyticICP/           # Core registration module
├── SmoothAdjustment/      # Analytic operations and deformation
├── Test/                  # Executable interface
├── data/                  # Sample datasets (2D & 3D)
├── experiment.ini         # Configuration file
├── Release/               # Compiled executables and runtime DLLs
└── README.md              # This file