Bilateral Texture Filtering

Overview

This repository hosts the MATLAB and C++ implementations of the Bilateral Texture Filter as presented in the paper by Cho, H., Lee, H., Kang, H., and Lee, S. (2014) titled "Bilateral texture filtering", published in ACM Transactions on Graphics, Vol. 33, No. 4, Article 128.

Contents

The repository includes:

• MATLAB implementation of the bilateral texture filter.
• C++ implementation of the same algorithm.
• Over 30 sample images in the images/ directory, demonstrating various textures. These images are named in the format name-k-iter.ext, where k is the optimal patch size and iter is the number of iterations for that image.

Getting Started

MATLAB Usage

1. Open the main.m file.
2. Set the input image I, patch size K, and number of iterations ITER.
3. Run the script by typing main in the MATLAB console.

C++ Installation and Usage

cd Cpp
mkdir build
cd build
cmake ..
make -j

Note: While the code is optimized to utilize all CPU threads, it is not real-time. The image compression4-5-15.png is the most time-consuming, averaging 12.4206 seconds over 10 runs on an i7-13700K CPU with 24 cores.