Computer Vision — Lab Portfolio

MATLAB implementations of classical computer vision algorithms, written from scratch as self-study for the Computer Vision course (code 77802) of the MSc in Robotics Engineering, DIBRIS — University of Genoa.

Each lab is a self-contained MATLAB project with its own entry-point script, input data, and a full LaTeX report (source + compiled PDF) discussing the method and the results.

Lab	Topic	Key algorithms
Lab2	Image filtering and the Fourier transform	Gaussian & salt-and-pepper noise models, moving-average / low-pass Gaussian / median filtering, linear filter design, 2-D FFT analysis
Lab5	Segmentation and feature detection	Normalized cross-correlation template matching, HSV colour-based segmentation, Harris corner detector
Lab6	Epipolar geometry	8-point algorithm, normalized (Hartley) 8-point algorithm, RANSAC fundamental-matrix estimation, SIFT/NCC feature matching, epipolar line and epipole visualization
Lab7	Motion analysis and tracking	Lucas–Kanade optical flow, static vs. running-average background modelling, change detection, blob-association tracker

Tech stack

• MATLAB throughout (no other language in the repository).
• Toolboxes used: Image Processing Toolbox (imhist, medfilt2, fspecial, normxcorr2, bwlabel, regionprops, rgb2hsv, …) and Computer Vision Toolbox (detectSIFTFeatures, extractFeatures, VideoReader). Lab 6 therefore needs a MATLAB release in which detectSIFTFeatures is available.
• Reports are LaTeX (article class, biblatex/biber); each Report/ directory contains the sources plus the compiled main.pdf.

Repository layout

Lab2/    Image filtering and Fourier transform
Lab5/    NCC-based segmentation and Harris corner detection
Lab6/    Fundamental matrix estimation
Lab7/    Motion analysis and tracking

Every lab directory holds its MATLAB sources, the input images or videos it needs, a Report/ folder with the LaTeX write-up, and a README.md describing what is implemented and how to run it.

Running a lab

git clone https://github.com/amirmat98/ComputerVision.git
cd ComputerVision/Lab2      # or Lab5, Lab6, Lab7

Open MATLAB, set the working directory to that lab folder (the scripts read their input files with relative paths), and run the lab's entry-point script:

Lab	Entry point
Lab2	main.m
Lab5	Main.m
Lab6	main_part1.m, main_part2.m
Lab7	main.m

Author

AmirMahdi Matin — MSc Robotics Engineering, University of Genoa. Course held by Prof. Nicoletta Noceti and Prof. Fabio Solari.

License

MIT