The objective of this project is to implement from scratch in CUDA C++ various image processing algorithms. A Cpu and a Gpu version of the following algorithms is implemented and commented:

• Canny Edge Detection
• Non Local-Means De-Noising
• K-Nearest Neighbors De-Noising
• Convolution Blurring
• Pixelize

We benchmarked the Gpu and Cpu version.

Make sure you have a CUDA capable GPU and install cudatoolkit for your OS. Then run:

    cd src/gpu
    mkdir build && cd build
    cmake ..
    make

Detects the edges of an image.

Usage:

    ./main <image_path> edge_detect

Removes the grain of an image.
Benchmark:

• Cpu:

• Gpu:

Usage:

    ./main <image_path> nlm <conv_size> <hyper_param>

Removes the noise of an image using the KNN algorithm.
Benchmark:

• Cpu:

• Gpu:

Usage:

    ./main <image_path> nlm <conv_size> <block_radius> <weight_decay>

Blurs an image using the convolution operator.
Usage:

    ./main <image_path> conv <conv_size>
    ./main <image_path> shared_conv <conv_size>

Use shared_conv for an optimized version using shared memory.

Pixelizes an image.
Usage

    ./main <image_path> pixelize <conv_size>