图像压缩格式详解：JPEG, PNG, GIF, XBM, BMP

需积分: 10 106 浏览量更新于2024-07-28 收藏 3.01MB PDF 举报

"Addison Wesley - Compressed Image File Formats-JPEG, PNG, GIF, XBM, BMP (Miano)" 是一本由John Miano编著的书籍，专注于介绍网络上常见的压缩图像文件格式，包括JPEG、PNG、GIF、XBM和BMP。本书旨在指导读者理解和掌握这些图形文件在Web上的应用。 JPEG（Joint Photographic Experts Group）是一种广泛使用的有损压缩格式，适用于照片和连续色调的图像。它通过丢弃人眼难以察觉的图像信息来实现高比例的压缩，从而降低文件大小，适合网络传输和存储。 PNG（Portable Network Graphics）是一种无损压缩格式，支持透明度和24位颜色。PNG用于需要高质量图像和透明效果的场合，例如网页设计和图标制作。 GIF（Graphics Interchange Format）是另一种流行格式，支持有限的256色调色板和透明度。GIF特别适合动画，但因其有损压缩算法和颜色限制，对于复杂图像和照片质量不如JPEG。 XBM（X BitMap）是一种简单的黑白图像格式，主要用于创建二进制位图，如计算机图标或软件界面元素。XBM文件以C语言源代码的形式存储，可以直接被编程语言处理。 BMP（Bitmap）是Windows操作系统中的标准位图格式，不进行压缩，因此文件通常较大。BMP支持丰富的色彩，但不适合网络传输，更适合本地存储和编辑。这本书可能涵盖了图像压缩的基本原理，每种格式的压缩算法，以及如何在不同情况下选择合适的文件格式。此外，可能还讨论了色彩管理、文件大小优化、跨平台兼容性以及与Web浏览器和图形处理软件的交互等主题。读者可以通过这本书深入了解图像文件的内部结构，学习如何有效地使用和处理这些格式，以适应不同的应用场景和需求。

Introduction

images, 8 is the most commonly used sample precision. However, 1, 2, 4, 12, and

16 are also common.

Integer component values can range from 0 to 2

Sample Precision

- 1. To convert

from the normalized real number representation of a color to the integer repre-

sentation you simply multiply by 2

Sample Precision

- 1.

On Windows the sample precision is almost always 8 bits, so the operating

system (but not necessarily the underlying hardware) recognizes 256 different

shades of each primary color. Other systems may use a larger or smaller sample

precision.

Grayscale

Some display devices, such as laser printers, cannot display colors at all but

rather shades of gray. These are known as grayscale devices. Shades of gray can

be represented by a single component in the range 0 (black) to 2

Sample Precision

- 1

(white). In Figure 1.5 you can see that shades of gray occur in the RGB model

along the line where R=G=B.

YCbCr Color Model

RGB is not the only color model in use. At one time the HSB (Hue-Saturation-

Brightness) color model was commonly used in computer systems and still is

used by some image processing applications. JPEG images are almost always

stored using a three-component color space known as YCbCr. The Y, or lumi-

nance, component represents the intensity of the image. Cb and Cr are the

chrominance components. Cb specifies the blueness of the image and Cr gives

the redness.

The YCbCr color model is similar to the one used in television sets that

allows color images to be compatible with black and white sets. In the YCbCr

color model, the Y component on its own is a grayscale representation of the

color image.

The relation between the YCbCr and RGB models as used in JPEG is repre-

sented in Equation 1.1.

Figure 1.6 shows a color image that has been separated into its Y, Cb, and Cr

components. You can see that the Y component contributes the most information

to the image. Unlike the RGB color model, where all components are roughly

Equation 1.1

YCbCr/RGB

Color space

Conversion

Y = 0.299R + 0.587G + 0.114B

Cb = -0.1687R - 0.3313G + 0.5B + 2

Sample Precision/2

Cr = 0.5R - 0.4187G - 0.0813B + 2

Sample Precision/2

R = Y + 1.402Cr

G = Y - 0.34414(Cb-2

Sample Precision/2

)-0.71414(Cr-2

Sample Precision/2

)

B = Y + 1.722(Cb - 2

Sample Precision/2

)

Introduction

equal, YCbCr concentrates the most important information in one component.

This makes it possible to get greater compression by including more data from

the Y component than from the Cb and Cr components.

CMYK Color Model

One other color model worth mentioning at this point is a four-component model

known as CMYK (cyan, magenta, yellow, black), which is frequently used in

color printing. Most printing is done on white paper with ink added to create

darker colors. This is the opposite of what happens on a computer monitor. The

CMYK colorspace follows the printing model. The components represent the

four inks commonly used in color printing.

The color models we have looked at so far are known as additive, which

means the components add light to the image. The higher the component values

are, the closer the color is to white. However, in CMYK, larger component val-

ues represent colors close to black. This is known as subtractive. Cyan, magenta,

and yellow are the complements for red, blue, and green. A pure cyan surface

absorbs all the red light directed at it. If yellow and magenta inks are combined,

they absorb the green and blue light, which results in red. Cyan, magenta, and

yellow combine to absorb all light, resulting in black—in theory, anyway.

In practice, cyan, magenta, and yellow inks do not combine on a white piece

of paper to produce a pure black. Even if you could get a good shade of black by

combining colors, it would require three times as much ink as simply using black

alone. Since most printing is done in black anyway, it makes sense to use black

ink to produce black and shades of gray.

On a computer monitor, the relationship between RGB and CMYK can be

approximated as shown in Equation 1.2.

Equation 1.2

CMYK/RGB

Colorspace

Conversion

K = (2

Sample Precision/2

- 1) - MAX(R,G,B)

C = (2

SamplePrecision/2

- 1) - R - K

Y = (2

Sample Precision/2

- 1) - G - K

M = (2

Sample Precision/2

- 1)-B-K

R = (2

Sample Precision/2

- 1) - K - C

G = (2

Sample Precision/2

- 1) - K - Y

B = (2

Sample Precision/2

- 1)-K-M

When the C, M, and Y component values are equal, the color is a shade of

gray. Notice how the conversion from RGB to CMYK replaces cyan, magenta,

and yellow ink with shades of gray produced by black ink. The result of this sub-

stitution is that at least one of the CMY components will always be zero if this

conversion process is followed exactly as shown here. The CMYK color model

does not require the value of one component to be zero—this is simply a result

True Color versus Palette

of converting from RGB. Applications that use the CMYK color model will

allow any combination of component values to give complete control over the

printed colors and allow for variations among various types of ink.

Another thing to notice about the CMYK color model is that there is not a

one-to-one mapping between it and RGB. Instead, multiple CMYK values map

to the same RGB value.

True Color versus Palette

The examples in this book assume that the output device uses the RGB color

model to display images and that each component is represented using 8 bits and

is in the range 0-255.

This is the color representation used by most personal

computer systems. Such a system can produce 16,777,216 (256

) distinct colors.

There are computer systems that use more bits to represent color, for example the

12-bit grayscale frequently used for medical images. Some image formats sup-

port data with more than 8 bits per component (12 for JPEG, 16 for PNG). For

the remainder of this discussion we are going to assume that you are working on

a system that uses 8 bits per component.

Two methods are commonly used to assign one of the possible colors to a

pixel. The simplest is to store the color value for each pixel in the compressed

data. For images with 24 bits per pixel, each pixel has a 3-byte color value asso-

ciated with it. Images that use 24 bits or more are called true color because over

the range of colors a monitor can display, 24 bits per pixel is the limit of color

differences that a human can distinguish.

The problem with 24-bit graphics is that while a system may be capable of

displaying 16,777,216 different colors, it may not be able to do so simultane-

ously. Older computers may not even have a video card capable of using a 24-bit

display mode. Newer computers may not have enough video memory to operate

in 24-bit mode at higher screen resolutions. A display on a personal computer set

at a resolution of 1024 × 768 pixels would require 2,359,296 (1024 × 768 × 3 =

2.25 MB) of video memory to display 24-bit images. If the computer had only

2 MB of video memory it could not display 24-bit images at this resolution but

could do so at a lower resolution of 800 x 600 (800 × 600 × 3 = 1.4 MB).

The solution devised to represent colors before the days of displays capable

of 24 bits per pixel was to define a color palette that selects a subset of the pos-

sible colors. Conceptually the palette is a 1-dimensional array of 3-byte ele-

ments that specify the color. Rather than directly specifying the color, each pixel

value is an index into the color palette. The most common size for a palette is

256 entries where each pixel value consists of 8 bits. Most computers today can

To avoid dealing too deeply with specific system implementations, this section contains some simplifi-

cations as to the behavior of display devices.

剩余265页未读，继续阅读

lhf_alex

粉丝: 0
资源: 1

图像压缩格式详解：JPEG, PNG, GIF, XBM, BMP

Compressed Image File Formats JPEG, PNG, GIF, XBM, BMP

Addison Wesley - 2001 - Knuth - The Art of Computer Programming Vol I II III IV.part5.rar of 6

Addison Wesley - C++ Network Programming

Addison Wesley -- LDAP programming with Java.pdf

Addison Wesley - PostgreSQL - Introduction and Concepts

Addison Wesley - Essential C++

Addison Wesley - Programming Ruby

Addison Wesley - Design Patterns

Addison Wesley - Continuous Delivery

Addison Wesley - OpenGL Distilled

最新资源