Support Normal Estimation

[wweqg]

Summary:

Normal Estimation Implementation

Type of change:

✨ Enhancement

Changes Made:

Implemented the normal estimation functionality to the project. The primary focus was on improving the normal estimation process and making the code more efficient, readable, and extensible. Below is a summary of the key changes made:

1. Parallelized k-Nearest Neighbors Search using KDTree:

   • Optimized the k-nearest neighbors search using the KDTree data structure.
   • Employed the Rayon library for parallel processing, reducing computation time significantly.

2. Covariance Matrix Computation:

   • Implemented the computation of covariance matrices for accurate normal estimation.
   • Explained the need for covariance matrices in capturing local surface structure.

3. Eigenvalue-Eigenvector Computation:

   • Computed eigenvalues and eigenvectors from covariance matrices.
   • Clarified why the first eigenvector with the smallest eigenvalue represents the normal direction.

4. Specialized Normal Orientation Propagation:

   • Addressed challenges related to normal orientation consistency.
   • Designed a specialized Breadth-First Search (BFS) algorithm to propagate normal orientation.

Proposed Commit Message:

Implement Normal Estimation Enhancements 

This commit introduces significant enhancements to the normal estimation process. 
Parallelization, covariance matrix computation, eigenvalue-eigenvector computation, 
and specialized normal orientation propagation have been added to improve accuracy 
and efficiency.