Fundamental 3D Computer Vision

This document presents assignments and solutions for the Fundamental 3D Computer Vision (CE-344) course at Sharif University of Technology. Each assignment focuses on fundamental concepts of 3D vision, including image processing, camera projections, and 3D reconstruction.

---

HW1 - Signal & Image Processing

This assignment includes four tasks, each focusing on key image processing techniques:

• 2D-DFT using np.fft.fft2
• Fourier Transform implementation from scratch
• Image smoothing techniques
• Color space conversion and noise removal

Visual Representations:

• 2D-DFT:
  
• Fourier Transform Implementation:
  
• Smoothing:
  
• Color Space Conversion (HSV & YCbCr):
  
• Salt & Pepper Noise Removal:
  

For more details, refer to the notebook.

---

HW2 - 3D Geometry

This assignment includes four main parts:

• Implementing the camera matrix and projection
• Determining the camera matrix for a given scenario
• Applying the camera matrix to a vector
• Modifying rotation (R) and translation (T) matrices

Visual Representations:

• Camera Matrix Example:
  
• Projection Results:
  
• Sequential Rotations:
  

For more details, refer to the notebook.

---

HW3 - Cameras and Projections

This assignment explores:

• Understanding rotations in 3D space (quaternions and rotation matrices)
• Estimating camera pose from 2D-3D correspondences
• Image mosaicing using homography

Visual Representations:

• Rotation Visualization (Quaternion):
  

• Camera Pose Estimation & 3D to 2D Projection:
  

• Image Mosaicing Process:

  Final Panorama Result:
  

For more details, refer to the notebook.

---

HW4 - 3D Reconstruction from Two Views

This assignment covers:

• Implementing the eight-point algorithm
• Implementing the normalized eight-point algorithm

Visual Representations:

• Input Images:
  
• Reconstruction Result:
  

---

Instructor ✍

Professor Shohreh Kasaei
Sharif University of Technology - Image Processing Lab (IPL)

For more details and assignments, visit the repository.