jordane95 / ASCIIArt

Converting an RGB image to its ASCII encoding

ASCIIArt

Converting an RGB signboard image to its ASCII encoding.

The whole pipeline is mainly composed of 3 modules, edge detection, ocr and ascii matching. For more detail, please read the documentation in each folder.

Example

Here is an example of our system.

This is the input image, downloaded from Internet

After processed by the 3 modules, we can get the final ASCII encoding result:

  | F       _ ,,, ,,, _   %      P      
  ! L       B| .   .l+@    2'  TC       
  !         F[(     []@    |,   L       
  ]         F|]     |]H    q"  'L       
  ]         L|]     []@    |_  j]       
  | L      ^F``"""""``"M   M==~%8      ,
n [W"""""""""^^     ^^"""""""""^"""""M"C
@ [@         ,`%  PSYANGJI.COM       | m
  ]Q        <~--*                    j p
  'q   E    -,-' TE`==JEP  @'' '` '    P
--;    @ lg>~Wy _,k_ _H@   E    -={    `
. ]    ] HM   ] '' -~,I@   "p ,.1TI  j  

Usage

1. Environment setput

   conda create -n asciiart
   conda activate asciiart
   pip install -r requirements.txt
   

2. put your image in /images/ folder

3. run the model to do ascii conversion via

   sh run.sh
   

4. your results will be saved in the /ascii/results/ folder

Structure

ASCIIArt
├── edge                                    // Edge detection 
│   ├── hed.py                              // Holistically-Nested Edge Detection
│   ├── canny.py                            // Canny
│   ├── README.md                           // Documentation of the edge detection module
│   ...
├── ocr                                     // OCR module
│   ├── ocr.py                              // PaddleOCR
│   ├── utils.py                            // Some useful fonctions
│   └── README.md                           // Documentation of the OCR module
├── theme_color                             // Theme color extraction module
│   ├── images                              // Images
│   │   ├── roi.jpg                         // input
│   │   └── theme_color.png                 // output
│   ├── color.py                            // KMeans color clustering code
│   └── README.md                           // Documentation of the color extraction module
├── ascii                                   // ASCII matching module
│   ├── fonts                               // Fonts
│   │   └── Menlo.ttc                       // Menlo monospace font       
│   ├── results                             // Converting results
│   │   ├── whole.txt                       // ASCII encoding of the whole image
│   │   ├── limit.txt                       // ASCII encoding of the ROI
│   │   ...                  
│   ├── ascii.py                            // Core code
│   ├── metrics.py                          // Different metrics for patch similarity
│   ├── search.py                           // Search of different segmentation configuration
│   ├── tone.py                             // Tone based ascii art
│   └── README.md                           // Documentation of the ASCII matching module
├── README.md                               // Documentation of the whole project
...

TODO

• Modularization

• Optimization