Real Time Image Processing

This repository contains the results of practical works for the Real time processing of conventional and non-conventional images with GPUs class in Jean Monnet University, lectured by Professor Philippe Colantoni. This course introduces basic and advanced techniques dedicated for General-Purpose processing on Graphics Processing Unit (GPGPU). It introduces the basic concepts of parallel programming and shows how to use the computing power of modern GPUs for conventional/non-conventional images processing in real time.

Content

Introduction to parallel programming; Introduction to General-purpose processing on graphics processing units (GPGPU): GPGPU with shaders, CUDA; Image processing with graphic shaders and compute shaders (application with WebGL for web applications); CUDA based image processing (application with OpenCV for native applications) Case of studies: Implementation of conventional color image/video processing methods, Implementation of non-conventional image processing methods

Practical Work 1: Shaders Based Image Processing

Practical Work 2: OpenMP and CUDA Based Image Processing

run from the directory cuda_and_openmp

sh compile.sh

if you want to run only for one method, you can see the commands for each method inside the compile.sh.

the implemented anaglyph methods equations were taken from this link.

to execute run these:

CUDA

Anaglyph

./imagecuda_a <<image path>> <<method number>>

Example: ./imagecuda_a flower_resized.jpg 2

Gaussian

./imagecuda_ip <<image path>> <<kernel size divided by 2>> <<sigma>>

Example: ./imagecuda_ip flower_resized.jpg 3 1.0

Denoising

./imagecuda_dn <<image path>> <<neighborhood size for the covariance matrix divided by 2>> <<factor ratio applied to determine the gaussian kernel size>>

Example: ./imagecuda_dn flower_resized.jpg 3 1.0

OpenMP

Anaglyph

./opencv-omp-anaglyph flower_resized.jpg 2

Gaussian

./opencv-omp-processing flower_resized.jpg 3 1.0

Denoising

./opencv-omp-denoising flower_resized.jpg 3 10000000.0

Acknowledgements

parts of the code borrowed from code given in class.