Description

Course Description

This certification course provides a comprehensive introduction to the principles, techniques, and applications of Digital Image Processing (DIP). Students will learn how images are acquired, represented, processed, enhanced, and analysed using both classical techniques and modern
AI-based approaches. The course combines theory, mathematics, and hands-on programming to establish a solid conceptual foundation and practical skills.
By the end of the course, students will be able to:

• Understand fundamental concepts of image formation, perception, and representation.
• Apply spatial and frequency domain techniques for image enhancement.
• Perform image segmentation, filtering, and restoration.
• Implement compression and morphological processing.
• Explore advanced applications, including medical imaging, satellite imagery, and facial recognition.
• Develop small projects using Python, OpenCV, and deep learning frameworks.

This course is delivered in a blended mode (online + offline) over six months with lectures, labs, quizzes, assignments, and a final capstone project.

Intended Audience

• Undergraduate students (Computer Science, Electronics, Electrical, IT, Biomedical, AI/ML, Data Science).
• Early postgraduate students or researchers looking for a foundation in DIP.
• Professionals seeking to upgrade their skills in computer vision and AI.

Prerequisite Knowledge

To ensure smooth learning, participants are expected to have:

1.  Mathematics:
   • Linear algebra (vectors, matrices, eigenvalues).
   • Probability & random
   • Basic calculus (differentiation, integration).
2.  Programming:
   • Basic Python programming (functions, loops, lists, NumPy arrays).
   • Exposure to libraries like Matplotlib, NumPy, and Pandas (preferred but not mandatory).
3. General Background:
   • Understanding of digital signals/systems is helpful, but not compulsory.
   • No prior knowledge of image processing or AI is

Course Highlights

• Duration: 6 months (~60 hours of lectures + 40 hours of labs/project).
• Mode: Blended (online lectures + offline hands-on sessions).
• Learning Style: Conceptual lectures, live coding demos, lab exercises, and project-based learning.
• Tools & Frameworks: Python, OpenCV, NumPy, scikit-image, TensorFlow/PyTorch (for advanced topics).
• Certification: Issued upon successful completion of quizzes, assignments, and final project.

Module 1: Introduction & Fundamentals (16 hrs)

Topics

• Applications of DIP (medical, satellite, biometrics, industry)
• Image sensing and acquisition (cameras, scanners, sensors)
• Visual perception basics (human eye, brightness adaptation, color vision)
• Digital image representation (pixels, intensity levels, bit-depth)
• Sampling & quantization
• Mathematical preliminaries:
• Linear algebra (vectors, matrices, eigenvalues, eigenvectors)
• Probability & random variables
• Convolution, correlation
• Image file formats: BMP, JPEG, PNG, TIFF, DICOM
• Color fundamentals: RGB, HSV, YCbCr, CMY

Learning Outcomes

• Understand the scope and applications of DIP.
• Explain image formation, acquisition, and representation.
• Revise math concepts necessary for DIP.
• Work with different image file formats and color models.

Labs / Hands-on

• Read, display, and save images using Python (OpenCV/Matplotlib).
• Convert color images to grayscale, binary, HSV.
• Explore pixel-level operations

Assessment

• Quiz 1 (math + basics)
• Lab report submission (image conversions)

Module 2: Image Enhancement (Spatial & Frequency Domain) (16 hrs)

Topics

• Point processing:
• Image negatives
• Log & power-law transformations
• Contrast stretching
• Histogram processing: equalization, specification
• Spatial filtering:
• Smoothing: averaging, Gaussian, median
• Sharpening: Laplacian, Sobel, Prewitt, Unsharp masking
• Discrete Fourier Transform (2D DFT)
• Properties of 2D DFT (translation, scaling, convolution theorem)
• Frequency domain filtering:
• Low-pass, high-pass, band-pass
• Homomorphic filtering

Learning Outcomes

• Perform intensity transformations and histogram equalization.
• Apply spatial filtering for smoothing and sharpening.
• Understand the Fourier transform and design filters in the frequency domain.

Labs / Hands-on

• Implement histogram equalization.
• Apply smoothing and sharpening filters.
• Fourier spectrum visualization.
• Apply frequency domain low-pass and high-pass filters.

Assessment

• Assignment 1 (implement 3 enhancement techniques).
• Quiz 2 (transformations, filtering).

Module 3: Image Transforms & Edge Detection (8 hrs)

Topics

Image transforms:
DCT (Discrete Cosine Transform)
Hadamard Transform
Karhunen-Loève Transform (KLT/PCA)
Edge detection:
Gradient-based (Sobel, Prewitt)
Laplacian of Gaussian (LoG)
Canny edge detector
Diffusion-based (isotropic, anisotropic)

Learning Outcomes

• Compare different image transforms and their properties.
• Detect edges using gradient, Laplacian, and Canny methods.
• Apply diffusion filtering for edge preservation.

Labs / Hands-on

• Apply DCT on image blocks and visualize coefficients.
• Implement Canny edge detection.
• Compare the results of Sobel, Prewitt, LoG, and Canny.

Assessment

Mini-project: Edge detection in medical/satellite images.

Module 4: Wavelets & Morphology (12 hrs)

Topics

• Wavelet transform concepts (scaling, mother wavelet, decomposition).
• Continuous & Discrete Wavelet Transform (CWT, DWT).
• 2D Wavelet decomposition.
•  Applications: edge detection, denoising, compression.
• Morphological image processing:
  • Erosion, dilation, opening, closing
  • Hit-or-miss transform
  • Skeletonization & pruning

Learning Outcomes

•  Understand multiresolution image representation.
• Apply wavelets for denoising and compression.
• Use morphology for binary image processing.

Labs / Hands-on

• Perform wavelet-based image compression.
• Apply DWT-based denoising.
• Implement erosion, dilation, opening, and closing on binary images.

Assessment

Assignment 2: Apply morphology to text/medical images

Module 5: Image Segmentation & Restoration (12 hrs)

Topics

•  Segmentation:
  • Thresholding (global, adaptive, Otsu’s method)
  • Region growing, split-and-merge
  • Watershed segmentation
  • Bayesian segmentation
• Image restoration:
  • Image degradation models
  • Noise models (Gaussian, Salt & Pepper, Poisson, Speckle)
  • Inverse filtering
  • Wiener filtering
  • Bayesian denoising

Learning Outcomes

• Implement thresholding and region-based segmentation methods.
• Handle noise and degradation models in images.
•  Apply filtering techniques for restoration.

Labs / Hands-on

• Implement Otsu thresholding.
•  Apply Watershed segmentation on real images.
• Restore noisy images using Wiener and Bayesian filters.

Assessment

Mid-term Exam (Modules 1–5).

Module 6: Image Compression & Color Processing (8 hrs)

Topics

Compression fundamentals:

• Redundancies (spatial, temporal, coding, psycho-visual)
• Lossless compression: Huffman, RLE, LZW
• Lossy compression: JPEG, JPEG2000, MPEG

Color image processing:

• Color models & transformations
• Smoothing & sharpening in color images
• Edge detection in color images

Learning Outcomes

• Explain redundancy in images.
• Apply Huffman coding and JPEG compression.
  Process and enhance color images.

Labs / Hands-on

Implement Huffman coding.
Compare JPEG compression at different quality levels.
Apply filtering on color images.

Assessment

Quiz 3 (Compression + Color Processing).

Module 7: Advanced Applications & AI in Image Processing (12 hrs)

Topics

Machine learning for image processing.
CNN basics (convolution, pooling, activation).
Transfer learning with pre-trained models (VGG, ResNet, MobileNet).

Applications:
Medical imaging (tumour detection, X-rays)
Satellite imagery (land use classification)
Face detection & recognition
OCR & document image analysis

Learning Outcomes

Understand the role of AI/ML in image processing.
Apply CNNs for simple classification tasks.
Use transfer learning for real-world applications.

Labs / Hands-on

Build a CNN to classify digits (MNIST).
Apply transfer learning to classify animals/objects.
Run a face detection pipeline using OpenCV + pre-trained model.

Assessment

Assignment 3: Apply transfer learning on a dataset (medical/satellite).

Module 8: Capstone Project & Evaluation (12 hrs)

Activities

Project ideation and proposal (Week 22).
Weekly mentoring sessions (Week 22–24).
Final presentations & viva (Week 24).

Sample Project Ideas

License plate recognition system.
Medical image enhancement (denoising, segmentation).
Satellite image segmentation for agriculture/urban areas.
Document scanner with OCR.
Face mask detection using deep learning.

Assessment

Final Project Evaluation (40%)
Viva & Presentation (10%)

Course Assessment Scheme (6 Months)

• Quizzes (3 × 5%) → 15%
• Assignments & Lab Reports → 15%
• Mini-projects (edge detection, morphology, CNN) → 10%
• Mid-term Exam (Modules 1–5) → 10%
• Final Project (report + code + presentation + viva) → 50%