Python图像处理库(PIL)

Python Imaging Library为您的python程序添加图像处理能力。这个库提供广泛的文件格式支持、高效的内部表示和相当强大的图像处理能力。

核心图像库是为快速访问几种基本像素格式图像设计的。它能为一般的图像处理工具提供一个可靠的基础。

这个Pythonic library可以

• 装载和保存多种格式文件 Loading and Saving images (diverse formats)  
  Python Imaging Library的最新版本可以识别并读取多数图像格式。写图像操作则有意地限制为最常用的交流和发布的格式。 convert images between different pixel representations 图像格式转换
• 缩放、裁剪、合成、变换和通道 Scaling, cropping, compositing, transforms, bands (the separate color channels e.g. alpha channel 半透明通道) 
  The library also supports image resizing, rotation and arbitrary affine transforms.提供图像缩放，旋转和任意仿射变换的支持。
• 图像增强 Image Enhancement 
  point operations,filtering with a set of built-in convolution kernels, and colour space conversions. 像素操作，内置convolution kernel的滤镜和颜色空间变换 
  adjust contrast, brightness, colour balance and sharpness using the ImageEnhance module
• 直方图、对比增强和统计分析 histogram,contrast enhancement,statistical analysis
  There‘s a histogram method allowing you to pull some statistics out of an image. This can be used for automatic contrast enhancement, and for global statistical analysis.提供histogram方法使您可以输出对某个图像的一些统计信息。这可以用于自动对比增强和全局统计分析。
• 绘制文字和基本图形 Drawing text and basic shapes
• 图像序列(动画) Image Sequences
  
  Supported sequence formats include FLI/FLC, GIF, and a few experimental formats. TIFF files can also contain more than one frame.
• 图像显示 Image Display
  The current release includes Tk PhotoImage and BitmapImage interfaces, as well as a Windows DIB interface that can be used with PythonWin. For X and Mac displays, you can use Jack Jansen‘s img library.最新发布版本包含Tk PhotoImage and BitmapImage 接口，Windows DIB 接口也可以随着PythonWin一起使用。而对了X 和Mac系统的显示，您可以使用Jack Jansen的图像库。
  For debugging, there‘s also a showmethod in the Unix version which calls xv to display the image. Debug版本中，于Unix也提供show方法调用xv来显示图像
• 图像打印 Postscript Printing
  print images, text and graphics on Postscript printers

几个应用的例子：

• 图像大小及格式转换 Image resizing or format conversions
• 添加保存/装载功能到程序 Adding save/load support into your application
• 图表 Chart generation (also see PIDDLE piddle.sf.net)
• 生成缩略图 Thumbnail generation
• 水印和图例 Watermarking or adding legend

基本概念 PIL API - Concepts

The size attribute is a 2-tuple containing width and height (in pixels).

通过图片对象的size属性可以得到图片的尺寸，结果这是一个二元组，包含水平和垂直方向上的像素数。

坐标 Coordinates

The graphics interface uses the same coordinate system as PIL itself, with (0, 0) in the upper left corner.

Pil采取左上角为(0,0)的坐标系统

角度 Angles
Angles are given in degrees. Zero degrees is in the +x (east) direction, and the angle increases

counter-clockwise,in the usual Cartesian convention. For example, angle 45 points northeast.

边界框 Bboxes（Bounding boxes）

A bounding box or bbox is a rectangle(矩形) in the image. It is defined by a 4-tuple, (x0, y0, x1, y1) where (x0,y0) is the top left (northwest) corner of the rectangle, and (x1, y1) is the bottom right (southeast) corner.
Generally, the area described by a bounding box will include point (x0, y0), but it will not include point (x1, y1) or the row and column of pixels containing point (x1, y1).
For example, drawing an ellipse inside the bounding box (0,0,5,10) will produce an ellipse 5 pixels
wide and 10 pixels high. The resulting ellipse will include pixel column 4 but not column 5, and will
also include pixel row 9 but not row 10.

由四元组(x0, y0, x1, y1)定义的一个矩形区域。区域包括左上角(x0,y0)，但不包括右下角(x1, y1)或者(x1, y1)所在行列的像素点。

通道 Bands – the separate color channels

An image band is a set of values, one per image pixel.

Monochrome(黑白) or grayscale images have one band;color images in the RGB system have three bands, CMYK images have four, and so on.

Photoshop users will recognize bands as similar to Photoshop channels.

一个图片可以包含一到多个数据通道，如果这些通道具有相同的维数和深度，Pil允许将这些通道进行叠加.

在RGB模式下，每个图片由三个通道叠加而成，每个模式下为一个灰度图，当有一个调色板来调色的时候，这三张灰度图的叠加即可合成3*8位(每个像素)的一个真彩图片。

pil库中，图片之间的模式(mode)可以转化。

>>> bands = im.split ()

>>> rIm = bands [0]

>>> aIm = bands [3]

>>> remadeImage = Image.merge (“RGBA”, (rIm, gIm, bIm, aIm))

>>> grayscaleIm = Image.open (“myAlpha.gif”)

>>> remadeImage = myImage.putalpha (grayscaleIm)  # replace the alpha band 

模式 Mode

the number and names of the bands in the image, and also the pixel type and depth.

Common modes are "L" (luminance) for greyscale images, "RGB" for true colour images, and "CMYK" for pre-press images.

1 (1-bit pixels, black and white, stored with one pixel per byte)

1位像素，黑和白，存成8位的像素
L (8-bit pixels, black and white)

8位像素，黑白
P (8-bit pixels, mapped to any other mode using a colour palette)

8位像素，使用调色板映射到任何其他模式
RGB (3x8-bit pixels, true colour)

3×8位像素，真彩
RGBA (4x8-bit pixels, true colour with transparency mask)

4×8位像素，真彩+透明通道
CMYK (4x8-bit pixels, colour separation)

4×8位像素，颜色隔离
YCbCr (3x8-bit pixels, colour video format)

3×8位像素，彩色视频格式
I (32-bit integer pixels)

32位整型像素
F (32-bit floating point pixels)

32位浮点型像素

调色板 Palette
mode("P")为每个像素定义具体的颜色值

颜色 Color – 32位数值、元组或字符串 specified as 32bit value, tuple, or string

• 单通道图像-像素数值32bit value

  For single-band images, the color is the pixel value.

  For “1”, “L”, and “I” images, use integers. 
  For “F” images, use integer or floating point values.
  For palette images (mode “P”), use integers as colour indexes.

  For example, in a mode "1" image, the color is a single integer, 0 for black, 1 for white. 
  For mode "L", it is an integer in the range [0,255], where 0 means black and 255 means white.

• 多通道图像-像素元组tuple
  For multi-band images, supply a tuple with one value per band. 
  For “RGB” images, use a 3-tuple containing integer values.The tuple (255,0,0) is pure red.
• 字符串常量 In 1.1.4 and later, you can also use RGB 3-tuples or colour names (see below). 颜色对照表
  • CSS风格的颜色字符串
    Hexadecimal color specifiers, given as “#rgb” or “#rrggbb”. For example, “#ff0000”specifies pure red.
  • RGB像素值 
    RGB functions, given as “rgb(red, green, blue)” where the colour values are integers in the range 0 to 255. Alternatively, the color values can be given as three percentages (0% to 100%). For example, “rgb(255,0,0)” and “rgb(100%,0%,0%)” both specify pure red.
  • HSL色彩模式
    Hue-Saturation-Lightness (HSL) functions, given as “hsl(hue, saturation%, lightness%)” 
    hue is the colour given as an angle between 0 and 360 (red=0, green=120, blue=240), 
    saturation is a value between 0% and 100% (gray=0%, full color=100%
    lightness is a value between 0% and 100% (black=0%, normal=50%, white=100%). 
    For example, “hsl(0,100%,50%)” is pure red.
    通过对色调(H)、饱和度(S)、亮度(L)三个颜色通道的变化以及它们相互之间的叠加来得到各式各样的颜色。
  • 常见HTML颜色名称
    Common HTML colour names. The ImageDraw provides some 140 standard colour names, based on the colors supported by the X Window system and most web browsers. Colour names are case insensitive, and may contain whitespace. 
    For example, “red” and “Red” both specify pure red.

滤镜 Resampling Filters
Some operations that reduce the number of pixels, such as creating a thumbnail, can use different filters to compute the new pixel values. These include:

在对图片的几何操作中可能会将多个输入像素映射到单个的输出像素（减少像素量，比如说创建缩略图），可以使用不同滤镜来计算新的像素。pil提供4种不同的采样滤镜(在目前的版本中，后续的版本可能支持更多)：

1 NEAREST最近 
Uses the value of the nearest pixel.
2 BILINEAR双线性
Uses linear interpolation over a 2x2 set of adjacent pixels.
3 BICUBIC双三次
Uses cubic interpolation over a 4x4 set of pixels.
4 ANTIALIAS平滑
Neighboring pixels are resampled to find the new pixel value.

例如，Image对象有三种使用filter的方法

1 resize(size,filter=None)
返回指定大小的一个新图像对象，可以使用一个filter参数来指明内插方式，默认采用NEAREST

>>> smallIm = im.resize ( (128, 128), Image.ANTIALIAS)

2 thumbnail(size,filter=None)

有两个特点：
1 改变原图 Modifies in-place
2 保持比例 aspect ratio(height : width)) -- size为(400,150)的im, im.thumbnail((40,40))后变为(40,15)3 filter(name)
Return a copy of the image filtered through a named image enhancement filter. 
有专门的ImageFilter模块来支持Image enhancement。

字体 Fonts 

PIL可以使用bitmap fonts或者OpenType/TrueType fonts.

Bitmap fonts are stored in PIL’s own format, where each font typically consists of a two files, one named .pil and the other usually named .pbm. The former contains font metrics, the latter raster data.

To load a bitmap font, use the load functions in the ImageFont module.

To load a OpenType/TrueType font, use the truetype function in the ImageFont module. Note that this function depends on third-party libraries, and may not available in all PIL builds.(IronPIL)

To load a built-in font, use the Font constructor in the ImageFont module.

有专门的ImageFont模块来支持ImageDraw模块中.text()方法中的字体选项。

文件格式 Format

The format attribute identifies the source of an image. If the image was not read from a
file, it is set to None.

文件类型码(Type列)出现在Image.format属性和Image.save()方法中：

>>> im.save (“vacation.png”)  #OR
>>> im.save (“vacation.png”, "PNG”)

常用操作

缩略图 thumbnail(size,filter=None)  
Modifies in-place,Preserves aspect ratio
>>> myImage.thumbnail ((128, 128), Image.ANTIALIAS)
剪切 crop(bbox)

>>> bounds = (100, 100, 400, 400)
>>> cutoutIm = myImage.crop (bounds)
粘贴 paste(i2,where,mask=None)/paste(color,box=None,mask=None)

旋转 rotate(theta)
rotated around its center

翻转旋转 transpose(method)
ROTATE_90/180/270(clockwise), FLIP_TOP_BOTTOM(horizontal), FLIP_RIGHT_LEFT(vertical)
>>> fixedIm = myImage.transpose (ROTATE_90)

The Image Module

• a class with the same name which is used to represent a PIL image.
• The module also provides a number of factory functions(including functions to load images from files, and to create new images)

图像对象 Image – from file or newly created

所有的图片操作必须有一个操作对象，例如Pil提供open(filename)进行这个过程，此后，一切关于图片的操作均基于这个对象。有以下几种创建image对象的方式：

1 Image.open(f)

>>> import Image

>>>

>>> Im = Image.open("lena.jpg")

>>> print Im.mode,Im.size,Im.format

RGB (256, 256) JPEG

>>> Im.show()

如果文件不能打开，会抛出IOError异常。

可以查看image对象的format,mode,size,palette,info几个属性。

调用im.show()会在图片查看工具中显示当前操作的image对象。

标准版本的show方法的实现不太高效，因为它先把image保存到一个临时文件，然后调用xy工具来显示图像。如果你没有安装xy，那么它就无法工作了。不过如果它可以工作，倒还是非常方便用来debug和测试。

2 Image.new(mode,size,color=None)

color的默认值是黑色，这里我们新建一个红色的图像。

>>> newIm = Image.new (“RGBA”, (640, 480), (255, 0, 0)) #新建一个image对象creating images from scratch

3 Image.blend(i1,i2,a)  -- (p1 x (1 - a) + p2 x a)

选一张灰度图(L)做背景，和雷娜图(RGB)做blend操作

>>> Im2 = Image.open("background.jpg").convert(Im.mode)

>>> Im2 = Im2.resize(Im.size)

>>> Im2.show()

>>>

>>> img = Image.blend(Im,Im2,0.2)

>>> img.show()

操作完毕后save(filename)用以保存这个临时的image对象img到硬盘。

4 Image.composite(i1,i2,mask)   --equal-sized images i1 ,i2 and mask("1", "L", or "RGBA") (p1 x (1 - m) + p2 x m)

5 Image.eval(f,i)  -- applying a function f to each pixel of image i

6 Image.merge(mode,bandList)  --Creates a multi-band image from a sequence of single-band images of equal size

以下是Image对象的全部方法：

The ImageDraw Module

• create new images,
• annotate or retouch existing images, and to generate graphics on the fly for web use.

For a more advanced drawing library for PIL, see The aggdraw Module.

创建绘画对象 ImageDraw module creates drawing surface for image

import Image, ImageDraw

im = Image.open(“vacation.jpeg") 

drawSurface = ImageDraw.Draw(im)

基本绘画操作 Basic methods of drawing surface

• 弧/弦/扇形 chord arc pieslice (bbox, strtAng, endAng)
• 椭圆 ellipse (bbox)
• 线段/多段线 line (L)  draw.line(((60,60),(90,60), (90,90), (60,90), (60,60))) #draw a square
• 点 point (xy)  #单像素点很小看不清,实际中可用实心小圆代替
• 多边形 polygon (L) draw.polygon([(60,60), (90,60), (90,90), (60,90)]) #draw a square
• 矩形 rectangle (bbox)       # first coord属于矩形, second coord不属于
• 文字 text(xy,message,font=None) 绘制文字message，文本区域左上角坐标为xy
        drawable.text((10, 10), "Hello", fill=(255,0,0), font=None)
• 文字大小 textsize(message,font=None)  给定文字message，返回所占像素(width,height)

可选参数 Common optional args for these methods

• fill=fillColor
• outline=outlineColor

矢量字体支持 TrueType Font support

import ImageFont

ttFont = ImageFont.truetype (“arial.ttf”, 16)

drawable.text ((10, 10), “Hello”, fill=(255,0,0), font=ttFont)

例子：Draw a Grey Cross Over an Image

import Image, ImageDraw im = Image.open("lena.pgm") # Creates an object that can be used to draw in the given image.
draw = ImageDraw.Draw(im) # draw.line(xy, options) => Draws a line between the coordinates in the xy list.

# The coordinate list can be any sequence object containing either 2-tuples [ (x, y), ... ] # or numeric values [ x, y, ... ]. # The fill option gives the color to use for the line.
draw.line((0, 0) + im.size, fill=128) draw.line((0, im.size[1], im.size[0], 0), fill=128) del draw # write to stdout
im.save(sys.stdout, "PNG")

The ImageChops module

These can be used for various purposes, including special effects 特殊效果, image compositions 图像合成, algorithmic painting 算法绘画, and more.

At this time, channel operations are only implemented for 8-bit images (e.g. "L" and "RGB").

例子:比较两幅图像

Exact Comparison：

The quickest way to determine if two images have exactly the same contents is to get the difference between the two images, and then calculate the bounding box of the non-zero regions in this image. If the images are identical, all pixels in the difference image are zero, and the bounding box function returns None.

import ImageChops def equal(im1, im2): return ImageChops.difference(im1, im2).getbbox() is None

To get a measure of how similar two images are, you can calculate the root-mean-square (RMS) value of the difference between the images. If the images are exactly identical, this value is zero. The following function uses the difference function, and then calculates the RMS value from the histogram of the resulting image.

RMS Difference：

To get a measure of how similar two images are, you can calculate the root-mean-square (RMS) value of the difference between the images. If the images are exactly identical, this value is zero. The following function uses the difference function, and then calculates the RMS value from the histogram of the resulting image.

# Example: File: imagediff.py

import ImageChops import math, operator def rmsdiff(im1, im2): "Calculate the root-mean-square difference between two images" h = ImageChops.difference(im1, im2).histogram() # calculate rms
    return math.sqrt(reduce(operator.add, map(lambda h, i: h*(i**2), h, range(256)) ) / (float(im1.size[0]) * im1.size[1]))