Goal-Line Technology in Soccer

This repository contains an undergraduate final-year MATLAB project that explores a simple computer-vision approach for deciding whether a soccer ball has crossed the goal line.

The project is best read as a prototype/demo rather than a production-ready goal-line technology system. It combines:

• goal-line detection,
• ball detection with Laplacian of Gaussian blob detection,
• geometric distance checks between the detected ball and goal line,
• frame annotation, and
• output video generation.

Project Overview

The main pipeline lives in finaltemp.m.

At a high level, it:

1. Detects or receives the goal-line endpoints through line1().
2. Iterates through extracted .jpg video frames.
3. Reads each frame and detects candidate ball blobs.
4. Chooses the most likely ball blob.
5. Computes the ball center/radius relative to the goal line.
6. Marks each frame as goal or no-goal.
7. Writes annotated frames into goal.avi.

The GUI entry point is guif.m, which appears to be generated with MATLAB GUIDE and wires buttons to the video/frame conversion, processing, and playback steps.

Repository Contents

File	Purpose
guif.m	MATLAB GUIDE GUI callbacks for running the demo pipeline and playing the generated result.
finaltemp.m	Main goal-line decision pipeline. Processes frames, detects the ball, checks goal/no-goal, and writes output video.
detect_blobs.m	Detects blobs by searching for local maxima in a Laplacian of Gaussian scale space.
create_scale_space.m	Builds the scale-space representation used by blob detection.
create_gaussian.m	Helper for generating a Gaussian kernel.
create_circle.m	Generates circle coordinates used to draw the detected ball overlay.

Algorithm Notes

The ball detection code uses a scale-space blob detector:

• input frames are normalized,
• each frame is convolved with Laplacian of Gaussian filters across multiple scales,
• local maxima above a threshold are treated as candidate blobs,
• candidate blobs are pruned and the most likely ball position is selected.

The goal/no-goal decision is based on geometry:

• the goal line is represented by two endpoints,
• the ball is represented by center coordinates and an estimated radius,
• the distance between the ball center and goal line is computed,
• the decision compares the ball position and radius against the line.

Requirements

This project was written for older MATLAB versions, around the MATLAB R2012a era.

You will likely need:

• MATLAB
• Image Processing Toolbox
• Video processing support, including VideoWriter

Some functions in the code are legacy MATLAB APIs, for example wavread. On newer MATLAB versions, replace wavread with audioread.

Expected Local Inputs

The original project expected several local files and helper functions that are not currently included in this repository:

• line1.m for goal-line endpoint detection
• vid.m for converting video into frames
• read_image.m
• prune_blobs.m
• pad_image.m
• is_maximum.m
• input .jpg frames
• bg.png for the GUI background
• goalsound
• generated/output folders such as images/

Because of those missing local assets, the repository is most useful today as readable project code and algorithm reference. To run it end-to-end, restore the missing helpers/assets or replace those calls with your own equivalents.

Running The Demo

If you have restored the expected input files and helper functions:

1. Open MATLAB.

2. Add this repository to the MATLAB path.

3. Update the hardcoded paths in finaltemp.m:

   base_dir = 'C:\Program Files\MATLAB\R2012a\bin\dagasirr';
   alt_dir = 'C:\Program Files\MATLAB\R2012a\bin\dagasirr\images';

4. Run the GUI:

   guif

5. Or run the processing script directly:

   finaltemp

The script writes annotated frames and creates goal.avi.

Modernization Ideas

If you want to revive this project, useful next steps would be:

• remove hardcoded Windows paths and use paths relative to the repository,
• replace missing helper functions with documented implementations,
• include a tiny sample input video or a few test frames,
• update legacy MATLAB APIs such as wavread,
• split finaltemp.m into smaller functions,
• add comments around the goal/no-goal decision thresholds,
• add screenshots or a short GIF of the expected output.

License

MIT License. See LICENSE.