SF2568-project

File structure

.
├── README.md               
├── data                # The data dir for script to use
│   ├── in-image        # The original images
│   ├── in-bin          # The bin files from original images
│   ├── out-image       # The bin files after edge detection
│   └── out-bin         # New images after edge detection
├── edge-detection.c    
├── utils.py            # Translating original images to bin file
├── test.sh             # Shell script for testing
└── requirements.txt    # pip install -r [this file]

Pipeline Overview

Color Image -- (Python script) --> Gray Scale Imaging as 2D array in Binary File

Binary File (original image) -- Edge detection (in C) --> Binary File (new image only contains edge)

Binary File -- (Python script) --> Gray Scale Imaging (only contains edge)

User guide

1. First, prepare the image in any format as long as it is accepted by my Python library Pillow.
2. Use the following command to pre-process the image to the binary file.

   python3 utils.py img-to-bin [in image file] [out binary file]
   

3. Compile the source code edge-detection.c and put the option -lm at the end. For example, cc edge-detection.c -O3 -lm.
4. Run the executable. The executable takes two arguments: the input binary file and the output binary file. The example call on PDC will be,

   srun [executable] [input binary file] [output binary file]
   

5. Use the following command to post-process the binary file to the image.

   python3 utils.py img-to-bin [out binary file]  [in image file] [in image file]
   

   Note that the Python script always outputs a PNG image regardless of the argument's format to ensure lossless compression.

Scripts

1. test.sh, used as followed

   ./test.sh [image file without path and postfix] [image file postfix]

   This is intended to speed up the test, assuming you have already put the corresponding file in the directory /data/in-image/, and has mpicc (from openmpi) configured.

   Example call: I have put Southbank-3.jpg in the directory as /data/in-image/Southbank-3.jpg. Using this script will give you the output image in /data/out-image/Southbank-3.png.

   ./test.sh Southbank-3 jpg