前言

字符画这个话题，似乎早在贴吧时代就已经被玩烂了。在百度图片随便一搜索，就能够看到非常多。然后在这个时代，会编程的人越来越多（尤其是 MATLAB，Python 等非常适合图像处理的脚本语言），类似的教程更是数不胜数。
在这里插入图片描述

一、字符画的实现原理

字符画是一系列字符的组合，我们可以把字符看作是比较大块的像素，一个字符能表现一种颜色（暂且这么理解吧），字符的种类越多，可以表现的颜色也越多，图片也会更有层次感。

问题来了，我们是要转换一张彩色的图片，这么多的颜色，要怎么对应到单色的字符画上去？这里就要介绍灰度值的概念了。

灰度值：指黑白图像中点的颜色深度，范围一般从0到255，白色为255，黑色为0，故黑白图片也称灰度图像
我们可以使用灰度值公式将像素的 RGB 值映射到灰度值：

gray ＝ 0.2126 * r + 0.7152 * g + 0.0722 * b

这样就好办了，我们可以创建一个不重复的字符列表，灰度值小（暗）的用列表开头的符号，灰度值大（亮）的用列表末尾的符号。
在这里插入图片描述

二、黑白字符画实现代码

demo.py test.jpg -o outfile.txt --width 90 --height 90

代码如下（示例）：

from PIL import Image
import argparse

def get_char(r,g,b,a=256):
    if a == 0:
        return ' '
    gray = 0.2126 * r + 0.7152 * g + 0.0722 * b
    length = len(ascii_str)
    unit = 256/length
    return ascii_str[int(gray/unit)]

if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument('file')  # 需要设置输入文件
    parser.add_argument('-o', '--output')  # 输出文件
    parser.add_argument('--width', type=int, default=80)  # 输出字符画宽
    parser.add_argument('--height', type=int, default=80)  # 输出字符画高

    # 获取参数
    args = parser.parse_args()
    IMG = args.file
    WIDTH = args.width
    HEIGHT = args.height
    OUTPUT = args.output

    ascii_str = list("$@B%8&WM#*oahkbdpqwmZO0QLCJUYXzcvunxrjft/\|()1{}[]?-_+~<>i!lI;:,\"^`'. ")
    im = Image.open(IMG)
    im = im.resize((WIDTH,HEIGHT))
    txt = ''
    for i in range(HEIGHT):
        for j in range(WIDTH):
            txt += get_char(*im.getpixel((j,i))) # (r,g,b,a)
        txt += '\n'

    print(txt)
    #字符画输出到文件
    if OUTPUT:
        with open(OUTPUT,'w') as f:
            f.write(txt)
    else:
        with open("output.txt",'w') as f:
            f.write(txt)

三、彩色字符画生成

在这里插入图片描述

代码实现：

import numpy as np
import cv2
from PIL import Image, ImageFont, ImageDraw, ImageFilter
import random
from pathlib import Path
import time
from tqdm import tqdm

def color(
    input: str,
    output: str = None,
    rows: int = 100,
    alphabet='uppercase',
    background='origin7',
    out_height: int = None,
    scale: float = None,
):
    """output colorful text picture"""
    input_path = Path(input)
    # the original image
    origin = Image.open(input_path)
    width, height = origin.size
    print(f'input size: {origin.size}')
    # text amount of the output image
    text_rows = rows
    text_cols = round(width / (height / text_rows) * 1.25)  # char height-width ratio
    origin_ref_np = cv2.resize(
        np.array(origin), (text_cols, text_rows), interpolation=cv2.INTER_AREA
    )
    origin_ref = Image.fromarray(origin_ref_np)
    # font properties
    fontsize = 17
    font = ImageFont.truetype('courbd.ttf', fontsize)
    char_width = 8.88
    char_height = 11
    # output size depend on the rows and cols
    canvas_height = round(text_rows * char_height)
    canvas_width = round(text_cols * char_width)
    # a canvas used to draw texts on it
    canvas = get_background(background, origin, canvas_width, canvas_height)
    print(f'canvas size: {canvas.size}')
    # start drawing
    since = time.time()
    print(f'Start transforming {input_path.name}')
    draw = ImageDraw.Draw(canvas)
    charlist = get_alphabet(alphabet)
    length = len(charlist)

    for i in tqdm(range(text_cols)):
        for j in range(text_rows):
            x = round(char_width * i)
            y = round(char_height * j - 4)
            char = charlist[random.randint(0, length - 1)]
            color = origin_ref.getpixel((i, j))
            draw.text((x, y), char, fill=color, font=font)
    # resize the reproduct if necessary
    if out_height:  # height goes first
        canvas_height = out_height
        canvas_width = round(width * canvas_height / height)
        canvas = canvas.resize((canvas_width, canvas_height), Image.BICUBIC)
    elif scale:
        canvas_width = round(width * scale)
        canvas_height = round(height * scale)
        canvas = canvas.resize((canvas_width, canvas_height), Image.BICUBIC)
    # output filename
    if output:
        output_path = Path(output)
    else:
        output_path = input_path.with_name(
            f'{input_path.stem}_{canvas_width}x{canvas_height}_D{text_rows}_{background}.png'
        )
    canvas.save(output_path)

    print(f'Transformation completed. Saved as {output_path.name}.')
    print(f'Output image size: {canvas_width}x{canvas_height}')
    print(f'Text density: {text_cols}x{text_rows}')
    print(f'Elapsed time: {time.time() - since:.4} second(s)')