hanziart

KO | EN

gitlite — search

#javascript #python #hacktoberfest #react #ai #typescript #llm #go #golang #android #machine-learning #rust #deep-learning #linux

(★ 9)

hanzi version of asciiart

hanziart Latest Version Download

Download Latest Version (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Hanziart

Hanzi (Chinese characters) version of `asciiart`.

ASCII art uses monospaced keyboard characters, originally limited to those encoded by the ASCII standard, in place of pixels to draw raster images. The intensity of a given pixel is approximated by the visual density of the ASCII character.

Chinese characters, or *hanzi*, also known by their Japanese name *kanji*, present an even wider variety of characters to use for ASCII art.

For an example, see `apple40.txt`. Best viewed in a text editor with line-wrapping turned off.

As a proxy for their visual density, we shall use the *stroke count* of a character. Stroke count data are provided in [Unihan database](https://www.unicode.org/reports/tr38/) produced by the Unicode consortium.

## Source data

You will need to download [Unihan data (zip archive)](http://www.unicode.org/Public/UCD/latest/ucd/Unihan.zip) from Unicode. The file we need is `Unihan_DictionaryLikeData.txt` within the `Unihan.zip` archive.

## Dependencies

Written in Python3. Requires the following libraries: scikit-image (`skimage`), `scipy`, `matplotlib`, `numpy`, `argparse`, `csv`, `random`, `re`, `collections`.

## Usage

```bash
python3 hanziart.py --image photo.jpeg
python3 hanziart.py --image photo.jpeg --gradelevel 2 # Use characters up to grade level 2
python3 hanziart.py --help # Help message
```

The only required argument is the path to the image file. Assumes that `Unihan_DictionaryLikeData.txt` is in the current folder by default. Use `--help` option to see all options and their defaults.

## To do list

Some ideas for future development

* Measure the visual density of each character, instead of using stroke count as a proxy.
 * Match the visual pattern of a character to the image, e.g. characters with strong diagonal elements better approximate part of an image with diagonal texture.
 * Use n-grams from actual written text, to better simulate readable text.

hanziart

hanziart Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

hanziart

hanziart Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Download README