Image类

Pillow中最重要的类就是Image,该类存在于同名的模块中。可以通过以下几种方式实例化:从文件中读取图片,处理其他图片得到,或者直接创建一个图片。

使用Image模块中的open函数打开一张图片:

1
2
3
4
5
6
>>> from PIL import Image
>>> im = Image.open("lena.ppm")
如果打开成功,返回一个Image对象,可以通过对象属性检查文件内容
>>> from __future__ import print_function
>>> print(im.format, im.size, im.mode)
PPM (512, 512) RGB

format属性定义了图像的格式,如果图像不是从文件打开的,那么该属性值为None;size属性是一个tuple,表示图像的宽和高(单位为像素);mode属性为表示图像的模式,常用的模式为:L为灰度图,RGB为真彩色,CMYK为pre-press图像。

如果文件不能打开,则抛出IOError异常。

当有一个Image对象时,可以用Image类的各个方法进行处理和操作图像,例如显示图片:

1
>>> im.show()

ps:标准版本的show()方法不是很有效率,因为它先将图像保存为一个临时文件,然后使用xv进行显示。如果没有安装xv,该函数甚至不能工作。但是该方法非常便于debug和test。(windows中应该调用默认图片查看器打开)

读写图片

Pillow库支持相当多的图片格式。直接使用Image模块中的open()函数读取图片,而不必先处理图片的格式,Pillow库自动根据文件决定格式。

Image模块中的save()函数可以保存图片,除非你指定文件格式,那么文件名中的扩展名用来指定文件格式。

图片转成jpg格式

1
2
3
4
5
6
7
8
9
10
11
from __future__ import print_function
import os, sys
from PIL import Image
for infile in sys.argv[1:]:
    f, e = os.path.splitext(infile)
    outfile = + ".jpg"
    if infile != outfile:
        try:
            Image.open(infile).save(outfile)
        except IOError:
            print("cannot convert", infile)

save函数的第二个参数可以用来指定图片格式,如果文件名中没有给出一个标准的图像格式,那么第二个参数是必须的。

创建缩略图

1
2
3
4
5
6
7
8
9
10
11
12
13
from __future__ import print_function
import os, sys
from PIL import Image
size = (128128)
for infile in sys.argv[1:]:
    outfile = os.path.splitext(infile)[0+ ".thumbnail"
    if infile != outfile:
        try:
            im = Image.open(infile)
            im.thumbnail(size)
            im.save(outfile, "JPEG")
        except IOError:
            print("cannot create thumbnail for", infile)

必须指出的是除非必须,Pillow不会解码或raster数据。当你打开一个文件,Pillow通过文件头确定文件格式,大小,mode等数据,余下数据直到需要时才处理。

这意味着打开文件非常快,与文件大小和压缩格式无关。下面的程序用来快速确定图片属性:

确定图片属性

1
2
3
4
5
6
7
8
9
from __future__ import print_function
import sys
from PIL import Image
for infile in sys.argv[1:]:
    try:
        with Image.open(infile) as im:
            print(infile, im.format"%dx%d" % im.size, im.mode)
    except IOError:
        pass

裁剪、粘贴、与合并图片

Image类包含还多操作图片区域的方法。如crop()方法可以从图片中提取一个子矩形

从图片中复制子图像

1
2
3
box = im.copy() #直接复制图像
box = (100100400400)
region = im.crop(box)

区域由4-tuple决定,该tuple中信息为(left, upper, right, lower)。 Pillow左边系统的原点(0,0)为图片的左上角。坐标中的数字单位为像素点,所以上例中截取的图片大小为300*300像素^2。

处理子图,粘贴回原图

1
2
region = region.transpose(Image.ROTATE_180)
im.paste(region, box)

将子图paste回原图时,子图的region必须和给定box的region吻合。该region不能超过原图。而原图和region的mode不需要匹配,Pillow会自动处理。

另一个例子

1
2
3
4
5
6
7
8
9
10
11
12
Rolling an image
def roll(image, delta):
    "Roll an image sideways"
    image = image.copy() #复制图像
    xsize, ysize = image.size
    delta = delta % xsize
    if delta == 0return image
    part1 = image.crop((00, delta, ysize))
    part2 = image.crop((delta, 0, xsize, ysize))
    image.paste(part2, (00, xsize-delta, ysize))
    image.paste(part1, (xsize-delta, 0, xsize, ysize))
    return image

分离和合并通道

1
2
r, g, b = im.split()
im = Image.merge("RGB", (b, g, r))

对于单通道图片,split()返回图像本身。为了处理单通道图片,必须先将图片转成RGB。

几何变换

Image类有resize()、rotate()和transpose()、transform()方法进行几何变换。

简单几何变换

1
2
out = im.resize((128128))
out = im.rotate(45# 顺时针角度表示

置换图像

1
2
3
4
5
out = im.transpose(Image.FLIP_LEFT_RIGHT)
out = im.transpose(Image.FLIP_TOP_BOTTOM)
out = im.transpose(Image.ROTATE_90)
out = im.transpose(Image.ROTATE_180)
out = im.transpose(Image.ROTATE_270)

transpose()和象的rotate()没有性能差别。

更通用的图像变换方法可以使用transform()

模式转换

convert()方法

模式转换

1
im = Image.open('lena.ppm').convert('L')

图像增强

Filter

ImageFilter模块包含很多预定义的增强filters,通过filter()方法使用

应用filters

1
2
from PIL import ImageFilter
out = im.filter(ImageFilter.DETAIL)

像素点处理

point()方法通过一个函数或者查询表对图像中的像素点进行处理(例如对比度操作)。

像素点变换

1
2
# multiply each pixel by 1.2
out = im.point(lambda i: i * 1.2)

上述方法可以利用简单的表达式进行图像处理,通过组合point()和paste()还能选择性地处理图片的某一区域。

处理单独通道

1
2
3
4
5
6
7
8
9
10
11
# split the image into individual bands
source = im.split()
R, G, B = 012
# select regions where red is less than 100
mask = source[R].point(lambda i: i < 100 and 255)
# process the green band
out = source[G].point(lambda i: i * 0.7)
# paste the processed band back, but only where red was < 100
source[G].paste(out, None, mask)
# build a new multiband image
im = Image.merge(im.mode, source)

注意到创建mask的语句:

1
mask = source[R].point(lambda i: i < 100 and 255)

该句可以用下句表示

1
imout = im.point(lambda i: expression and 255)

如果expression为假则返回expression的值为0(因为and语句已经可以得出结果了),否则返回255。(mask参数用法:当为0时,保留当前值,255为使用paste进来的值,中间则用于transparency效果)

高级图片增强

对其他高级图片增强,应该使用ImageEnhance模块 。一旦有一个Image对象,应用ImageEnhance对象就能快速地进行设置。 可以使用以下方法调整对比度、亮度、色平衡和锐利度。

图像增强

1
2
3
from PIL import ImageEnhance
enh = ImageEnhance.Contrast(im)
enh.enhance(1.3).show("30% more contrast")

动态图

Pillow支持一些动态图片的格式如FLI/FLC,GIF和其他一些处于实验阶段的格式。TIFF文件同样可以包含数帧图像。

当读取动态图时,PIL自动读取动态图的第一帧,可以使用seek和tell方法读取不同帧。

1
2
3
4
5
6
7
8
9
from PIL import Image
im = Image.open("animation.gif")
im.seek(1# skip to the second frame
try:
    while 1:
        im.seek(im.tell()+1)
        # do something to im
except EOFError:
    pass # end of sequence

当读取到最后一帧时,Pillow抛出EOFError异常。

当前版本只允许seek到下一帧。为了倒回之前,必须重新打开文件。

或者可以使用下述迭代器类

动态图迭代器类

1
2
3
4
5
6
7
8
9
10
11
12
13
class ImageSequence:
    def __init__(self, im):
        self.im = im
    def __getitem__(self, ix):
        try:
            if ix:
                self.im.seek(ix)
            return self.im
        except EOFError:
            raise IndexError # end of sequence
for frame in ImageSequence(im):
    # ...do something to frame...
Postscript Printing

Pillow允许通过Postscript Printer在图片上添加images、text、graphics。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Drawing Postscript
from PIL import Image
from PIL import PSDraw
im = Image.open("lena.ppm")
title = "lena"
box = (1*722*727*7210*72# in points
ps = PSDraw.PSDraw() # default is sys.stdout
ps.begin_document(title)
# draw the image (75 dpi)
ps.image(box, im, 75)
ps.rectangle(box)
# draw centered title
ps.setfont("HelveticaNarrow-Bold"36)
w, h, b = ps.textsize(title)
ps.text((4*72-w/21*72-h), title)
ps.end_document()

ps:textsize不能用,有谁知道吗

更多读取图片方法

之前说到Image模块的open()函数已经足够日常使用。该函数的参数也可以是一个文件对象。

从string中读取

1
2
import StringIO
im = Image.open(StringIO.StringIO(buffer))

从tar文件中读取

1
2
3
from PIL import TarIO
fp = TarIO.TarIO("Imaging.tar""Imaging/test/lena.ppm")
im = Image.open(fp)

草稿模式

draft()方法允许在不读取文件内容的情况下尽可能(可能不会完全等于给定的参数)地将图片转成给定模式和大小,这在生成缩略图的时候非常有效(速度要求比质量高的场合)。

draft模式

1
2
3
4
5
from __future__ import print_function
im = Image.open(file)
print("original =", im.mode, im.size)
im.draft("L", (100100))
print("draft =", im.mode, im.size)

www.qytang.com/
http://www.qytang.com/cn/list/29/
http://www.qytang.com/cn/list/28/610.htm
http://www.qytang.com/cn/list/28/595.htm
http://www.qytang.com/cn/list/28/583.htm
http://www.qytang.com/cn/list/28/582.htm
http://www.qytang.com/cn/list/28/576.htm
http://www.qytang.com/cn/list/28/523.htm
http://www.qytang.com/cn/list/28/499.htm
http://www.qytang.com/cn/list/28/488.htm
http://www.qytang.com/cn/list/28/466.htm
http://www.qytang.com/cn/list/28/463.htm
http://www.qytang.com/cn/list/28/458.htm
http://www.qytang.com/cn/list/28/455.htm
http://www.qytang.com/cn/list/28/447.htm

Python图像处理库:Pillow 初级教程-乾颐堂的更多相关文章

  1. Python图像处理库Pillow入门

    http://python.jobbole.com/84956/ Pillow是Python里的图像处理库(PIL:Python Image Library),提供了了广泛的文件格式支持,强大的图像处 ...

  2. Python图像处理库Pillow常用使用方法

    PIL(Python Imaging Library)是Python一个强大方便的图像处理库,只支持到Python2.7.Pillow是PIL的一个派生分支,在Python3中用Pillow代替PIL ...

  3. python图像处理库Pillow基本使用方法

    安装pillow pillow的文档页面,documentation of Pillow 生成一个有单一颜色的图像 from PIL import Image, ImageDraw img = Ima ...

  4. 高性能python编程之协程(stackless)-乾颐堂

    我们都知道并发(不是并行)编程目前有四种方式,多进程,多线程,异步,和协程. 多进程编程在python中有类似C的os.fork,当然还有更高层封装的multiprocessing标准库,在之前写过的 ...

  5. Python原始套接字编程-乾颐堂

    在实验中需要自己构造单独的HTTP数据报文,而使用SOCK_STREAM进行发送数据包,需要进行完整的TCP交互. 因此想使用原始套接字进行编程,直接构造数据包,并在IP层进行发送,即采用SOCK_R ...

  6. python 开发简单的聊天工具-乾颐堂

    python 太强大了,以至于它什么都可以做,哈哈,开个玩笑.但是今天要讲的真的是一个非常神奇的应用. 使用python写一个聊天工具 其实大家平时用的QQ类似的聊天工具,也是使用socket进行聊天 ...

  7. python中fork()函数生成子进程分析-乾颐堂

    python的os module中有fork()函数用于生成子进程,生成的子进程是父进程的镜像,但是它们有各自的地址空间,子进程复制一份父进程内存给自己,两个进程之 间的执行是相互独立的,其执行顺序可 ...

  8. python第三方库-图像处理库pillow

    python图像处理库pillow 安装 pip install pillow 使用 导入 from PIL import pillow 读取图像 picture = Image.open('test ...

  9. Python图像处理库:Pillow 初级教程

    Python图像处理库:Pillow 初级教程 2014-09-14 翻译 http://pillow.readthedocs.org/en/latest/handbook/tutorial.html ...

随机推荐

  1. 通过yum升级gcc/g++至版本4.8.2

    最近在坐一个日期处理的问题,需要安装sxtwl模块,但是gcc版本4.7死活也安装不上,最后测试到只能升级gcc版本到4.8才得以解决 [root@123 bin]# gcc -vUsing buil ...

  2. web service初探

    概述:Web service是一个平台独立.低耦合的.自包含的.基于可编程的web的应用程序,可使用开放的XML(标准通用标记语言下的一个子集)标准来描述.发布.发现.协调和配置这些应用程序,用于开发 ...

  3. python+selenium+requests爬取我的博客粉丝的名称

    爬取目标 1.本次代码是在python2上运行通过的,python3的最需改2行代码,用到其它python模块 selenium 2.53.6 +firefox 44 BeautifulSoup re ...

  4. delphi RAD Studio新版本及路线图 及官方网站 官方 版本发布时间

    delphi  RAD Studio Berlin 10.1 主要是FireMonkey 移动开发的改动,VCL确实没有多大变化. http://docwiki.embarcadero.com/RAD ...

  5. Redis 安装 和 启动

    Redis下载官网 http://download.redis.io/releases/  本人下载了stable版 1:安装步骤 ># wget http://download.redis.i ...

  6. Mysql binlog二进制日志

    Mysql binlog日志有三种格式,分别为Statement,MiXED,以及ROW! 1.Statement:每一条会修改数据的实际原sql语句都会被记录在binlog中. 优点:不需要记录每一 ...

  7. SQL语言类别

    SQL语言主要分为四大类:数据查询语言DQL,数据操纵语言DML, 数据定义语言DDL,数据控制语言DCL. DQL (data query language) DML(data manipulati ...

  8. Spring MVC 接受的请求参数

    目录 1. 概述 2. 详解 2.1 处理查询参数 2.2 处理路径参数接受输入 2.3 处理表单 3. 补充内容 3.1 Ajax/JSON 输入 3.2 multipart参数 3.3 接收 he ...

  9. urllib 和urllib2 模块使用简例

    一.最简单的使用 import urllib,urllib2 response = urllib2.urlopen("https://www.baidu.com") print r ...

  10. 基元线程同步构造 AutoResetEvent和ManualResetEvent 线程同步

    在.Net多线程编程中,AutoResetEvent和ManualResetEvent这两个类经常用到, 他们的用法很类似,但也有区别.ManualResetEvent和AutoResetEvent都 ...