CSE 4227 · Digital Image Processing · AUST

Digital Image Processing, from zero

Plain-words lessons for every slide deck, past questions solved the way the exam wants them, and simulators you can step through.

How to use this site

  1. Learn: go through the lessons in order. Each one explains its slide deck in plain words, with real graphs, every past question solved in answer-script form, and step-by-step simulators wherever a method appears.
  2. Practise: use the trainers for the methods that are easiest to get wrong on paper.
  3. Revise: before the final, use the revision pages. The ultimate page puts every past question next to its answer.

The course is CSE 4227 Digital Image Processing at AUST. The final paper has 3 sections: Q1 is theory (CLO1), Q2 is maths in the spatial domain (CLO2) and Q3 is case studies (CLO3). Questions marked ★ in the lessons are the ones that keep coming back.

Lessons, one per slide deck

Deck 01–02 · Ch 1–2Introduction and digital image fundamentalsHuman eye, image formation, sampling and quantization, resolution, L = 2^k, M×N×k bits, zooming by nearest neighbour.Deck 03 · Ch 2Pixel relationships4-, 8- and m-neighbours, adjacency, paths, connected components, D4 / D8 / Euclidean distance.Deck 04 · Ch 3Intensity transformationsNegative, log, power-law (gamma), piecewise linear, contrast stretching, gray-level slicing, bit-plane slicing.Deck 05 · Ch 3Histogram processingHistograms and the 4 image types, sliding, stretching, equalization step by step, histogram matching.Deck 06 · Ch 3Smoothing spatial filtersCorrelation vs convolution, mean and weighted average, borders, median, min and max filters, noise.Deck 07 · Ch 3Sharpening spatial filters1st and 2nd derivatives, Laplacian derivation and masks, sharpening g = f ± ∇²f, Sobel and Prewitt gradients.Deck 08 · Ch 10Segmentation I: points, lines, edges, CannyPoint and line masks, gradient magnitude and angle, LoG and zero crossings, Canny NMS and hysteresis.Deck 09 · Ch 10Segmentation II: thresholding and regionsGlobal and iterative thresholding, region growing, split and merge.Deck 10 · Ch 11Chain code4- and 8-direction chain codes, first difference, normalising for start point and rotation, shape number.Deck 11 · Ch 9Morphology IStructuring elements, erosion, dilation, opening, closing, boundary extraction, region filling, counting coins.Deck 12 · Ch 9Morphology IIHit-or-miss transform, corners and end points, thinning, thickening.Deck 13 · Ch 8Image compressionRedundancy, compression ratio, entropy, Huffman, Golomb, arithmetic, LZW, run-length coding, RMSE.Deck 14 · Ch 6Colour image processingColour fundamentals, RGB cube, CMY(K), HSI and the conversions.

Revision

RevisionUltimate revision: every past questionEach past final and quiz question as the paper image, with its worked answer and simulator, grouped by chapter, with the pattern and formula wall.RevisionMega sheetCue words, every formula with a worked line, Q1 theory points, Q3 case answers and traps on one page.RevisionSegmentation fast trackThe minimum of decks 08–09 for the exam, with numerics.RevisionFall 2025 quiz answersAll 12 Fall 2025 quiz questions, answered in full with simulators.RevisionPast questions onlyEvery past question as the paper image, without answers, for self-testing.

Trainers

TrainerRegion growing trainerCircle the pixels that join, check yourself, and get the exam write-up table.TrainerLZW trainerFill the LZW dictionary table the way it is done on paper.TrainerChain code trainerWalk a boundary, write the chain code, first difference and shape number.

Fix or add something

Every page is generated from Python in the GitHub repository, and every number on a page is computed there. To fix a mistake or add a lesson, edit the generator in src/, run python build.py, and open a pull request. The site rebuilds itself when a change lands on main. The README explains the layout.

Content: CC BY-NC-SA 4.0 (share and adapt with credit, non-commercial, same license). Code: MIT. Past-paper question images belong to AUST and are shown for study.