Digital Image Processing and Applications Project_1

Overview

Developed as a project for the Digital Image Processing and Applications course at Harokopio University of Athens – Dept. of Informatics and Telematics, this project implements a series of exercises related to image processing, compression, filtering, segmentation, and optimization. The exercises explore key concepts in computer vision and image analysis, including quantization, Fourier transform, filtering, denoising, and object detection.

All implementations are done in Python using libraries such as NumPy, OpenCV, scikit-image, scikit-learn, and Matplotlib.

---

Exercises

Exercise 1: Quantization

• Task A: Implement a color quantizer in the RGB space using the K-means clustering algorithm.
• Task B: Generate quantized versions of the image flowers.jpg with 5, 20, 200, and 1000 color levels.
• Task C: Compute the Mean Squared Error (MSE) for each quantization level and discuss the results.

---

Exercise 2: Fourier Transform

• Task A: Compute the 2D Discrete Fourier Transform (DFT) of the grayscale image cornfield.jpg and display the magnitude and phase spectra.
• Task B: Replace the phase spectrum with its vertically symmetric version and recompute the Fourier spectra.
• Task C: Compute the Inverse Fourier Transform of the modified spectrum and display the resulting image. Provide commentary on the results.

---

Exercise 3: Filtering

• Task A: For the image girlface.jpg, design and apply a filter for object/feature detection.
• Task B: For the image fruits.jpg, design and apply a filter for edge detection of objects.
• Task C: For the image leaf.jpg, design and apply a filter to enhance the fine details of the objects.

For each case:

• Display the filtered image.
• Provide a detailed explanation of the filter design process (e.g., filter type, parameters).

---

Exercise 4: Optimization

• Images Used: lenna.jpg (reference image) and noisy versions: lenna-n1.jpg, lenna-n2.jpg, lenna-n3.jpg.
• Task A: Design and apply suitable filters for denoising each noisy version of the image. Display the best result for each case and explain the filter design choices.
• Task B: Compute the SSIM (Structural Similarity Index Measure) between the denoised images and the reference image (lenna.jpg) and discuss the results.

---

Exercise 5: Segmentation (Object Detection)

• Task A: Design, describe, and implement a segmentation algorithm for detecting objects in the grayscale image parking-lot.jpg. For each detected object, compute a bounding box.
• Task B: Display the original image with all bounding boxes drawn in a distinct color. Provide analysis of the results.

---

Exercise 6: Segmentation (Character Detection)

• Task A: Convert the color image book-cover.jpeg to grayscale and implement a segmentation algorithm for detecting characters/letters in the image.
• Task B: Display a binary segmentation mask where the foreground corresponds to detected characters and the background corresponds to all other pixels. Discuss the effectiveness of the segmentation.

---

Technologies Used

• Python 3
• NumPy – Matrix and numerical operations
• OpenCV – Image manipulation and filtering
• scikit-image – Image processing utilities
• scikit-learn – K-means clustering for quantization
• Matplotlib – Visualization
• tqdm – Progress bars during dataset/image processing

---

Results and Discussion

• Quantization shows that increasing the number of colors reduces error but at the cost of storage size.
• Fourier transform manipulation demonstrates the importance of the phase spectrum in image reconstruction.
• Filtering enables targeted enhancement such as edge detection, feature extraction, and detail sharpening.
• Optimization through denoising highlights trade-offs between removing noise and preserving fine image details, validated using SSIM.
• Segmentation tasks demonstrate object and character detection using thresholding, morphological operations, and bounding box extraction.

---

How to Run

1. Clone this repository:

   git clone <https://github.com/AthosExarchou/Digital-Image-Processing-and-Applications-Project_1.git>
   cd <Digital-Image-Processing-and-Applications-Project_1>

2. Install dependencies:

   pip install -r requirements.txt

3. Run each exercise script individually (examples):

   python exercise1_quantization.py
   python exercise2_fourier.py
   python exercise3_filtering.py
   python exercise4_optimization.py
   python exercise5_segmentation.py
   python exercise6_character_segmentation.py

---

Author

• Name: Exarchou Athos
• Student ID: it2022134
• Email: [email protected], [email protected]

License

This project is licensed under the MIT License.