Batch Grayscale Conversion of Aerial TIFF Images Using CUDA

A small utility that scans a folder of high‑resolution TIFF images, converts each pixel from RGBA to 8‑bit grayscale on the GPU, and writes out _gray.tiff files. Leverages CUDA for parallel pixel processing and libtiff for fast disk I/O, while C++17’s <filesystem> handles directory traversal.

tldr: gray_out folder contains result and images contain orignal images images used are from the assignment links

---

Table of Contents

• Features
• Prerequisites
• Directory Structure
• Building
• Running
• Example
• License

---

Features

• Batch processing: Automatically finds all .tif/.tiff files in an input folder.
• GPU‑accelerated: Converts RGBA to grayscale in parallel on the CUDA device.
• LibTIFF I/O: Fast, lossless reading/writing of TIFF images.
• Modern C++17: Uses <filesystem> for clean directory traversal.

---

Prerequisites

• CUDA Toolkit (10.x or newer)
• libtiff development headers and library
• A C++17‑capable host compiler (e.g., g++, clang++)
• make, bash

On Ubuntu/Debian, you can install dependencies via

sudo apt update
sudo apt install build-essential libtiff-dev cuda-toolkit make

Directory Structure . ├── convert_batch.cu # main batch + kernel implementation ├── convert_batch.hpp # declarations for batch functions & kernel ├── Makefile # build rules ├── run.sh # simple runner script └── images/ # (place your .tiff inputs here)

Building

From the project root, simply run:

make

This compiles convert_batch.cu into the convert_batch executable, linking against libtiff and the CUDA runtime.

If you ever need to clean up:

make clean

Running

You can invoke the tool directly:

./convert_batch <input_dir> <output_dir> [-t threadsPerBlock]

Or use the provided run.sh wrapper:

bash run.sh <input_dir> <output_dir> [threadsPerBlock]

<input_dir>: folder containing your .tif/.tiff files

<output_dir>: folder where *_gray.tiff files will be written (created if missing)

threadsPerBlock (optional): number of threads per CUDA block (default: 256)

Example

Place your aerial images in images/.

Build the project:

make

Run the batch conversion:

bash run.sh images/ gray_out/ 512

Check gray_out/ for your converted _gray.tiff files.

Also i have tried following as much as c++ guidelines as possible