用PIL将二进制字符串转换成图像

Question

所以，我要做的是让PIL创建一个基于二进制字符串的图像。 一些背景：

from PIL import Image

value = "0110100001100101011011000110110001101111"
vdiv = [value[i:i+8] for i in range(0, len(value), 8)]

^这将从二进制字符串创建字节列表。['01101000', '01100101',.....]

def diff(inp):
    if inp == '1':
        return (0,0,0)
    if inp == '0':
        return (255,255,255)
    else:
        pass

^这将为每个对应的位返回一个颜色元组，如果我调用：

for i in vdiv:
    for i2 in i:
        print diff(i2)

它将为列出的字节中的每个位打印每个颜色元组。(0,0,0) (0,0,0) (255,255,255)...

我想知道的是，如何让PIL创建像素与二进制字符串匹配的图像？

它应该是这样的。：

到目前为止我对皮尔的看法是：

img = Image.new( 'RGB', (8,len(vdiv)), "white")
pixels = img.load()

##
for x in range(img.size[0]):
    for y in range(img.size[1]):
        for i in vdiv:
            for i2 in i:
                pixels[x,y] = diff(i2) #Creates a black image. What do?
##

img.show()

是for x in range位让我抓狂。我迷路了。

Answer 1

您可以使用img.putdata：

import Image

value = "0110100001100101011011000110110001101111"

cmap = {'0': (255,255,255),
        '1': (0,0,0)}

data = [cmap[letter] for letter in value]
img = Image.new('RGB', (8, len(value)//8), "white")
img.putdata(data)
img.show()        

---

如果您有NumPy，可以使用Image.fromarray：

import Image
import numpy as np

value = "0110100001100101011011000110110001101111"

carr = np.array([(255,255,255), (0,0,0)], dtype='uint8')
data = carr[np.array(map(int, list(value)))].reshape(-1, 8, 3)
img = Image.fromarray(data, 'RGB')
img.save('/tmp/out.png', 'PNG')

但这次测试表明，使用putdata更快：

value = "0110100001100101011011000110110001101111"*10**5

def using_fromarray():
    carr = np.array([(255,255,255), (0,0,0)], dtype='uint8')
    data = carr[np.array(map(int, list(value)))].reshape(-1, 8, 3)
    img = Image.fromarray(data, 'RGB')
    return img

def using_putdata():
    cmap = {'0': (255,255,255),
            '1': (0,0,0)}

    data = [cmap[letter] for letter in value]
    img = Image.new('RGB', (8, len(value)//8), "white")
    img.putdata(data)
    return img

---

In [79]: %timeit using_fromarray()
1 loops, best of 3: 1.67 s per loop

In [80]: %timeit using_putdata()
1 loops, best of 3: 632 ms per loop