Live Fourier Transform of Camera Feed

This Python program captures images from a camera feed, applies a Fourier transform (FFT) using PyTorch for GPU acceleration and multicore processing, and displays the transformed image alongside the original.

This can be used to explain concepts such as the Fourier Transform, but also scattering and diffraction effects, if the lecturer shows printouts of the library of shapes and arrays in front of the camera.

Prerequisites

Ensure you have Python 3.12 or above installed.

Installation

Install dependencies with:

pip install -r requirements.txt

Usage

Run the program with:

python LiveFT.py

Press "q" to exit the application

Options

• -n, --numShots: Maximum number of images before program exits.
• -d, --camDevice: Camera device ID to use.
• -i, --imAvgs: Number of images to average for display and FFT processing.
• -y, --vScale: Vertical video scaling factor.
• -x, --hScale: Horizontal video scaling factor.
• -p, --downScale: Enable pyramidal downscaling for performance improvements.
• -k, --killCenterLines: Remove central lines from FFT image to improve dynamic range.
• -f, --figid: Name for the display window.
• -r, --rows: Number of central rows to use in video frame cropping.
• -c, --columns: Number of central columns to use in video frame cropping.

Notes

• This program uses PyTorch for FFT calculations, with GPU acceleration if available.
• Tested on MacOS and Linux environments.

License

This project is licensed under the Apache-2.0 license.