Electronic imaging—Guidance for selection of document image compression methods
Some standard content:
ICS 37.080
National Standardization Guidance Technical Document of the People's Republic of China GB/Z 19736--2005/IS0/TS 12033:2001 Electronic imaging
Guidance for selection of document image compression methods
Electronic innaging
Guidance for selection of document image compression methods(ISO/TS12033.2001.IDT)
2005-04-19 Issued
General Administration of Quality Supervision, Inspection and Quarantine of the People's Republic of China Administration of Standardization of the People's Republic of China
2005-10-01 Implementation
irkAoNikAca
GB/Z19736—2005/IS0/TS12033:2001 Due to the rapid increase in the application of compression technology, compression methods play an increasingly important factor in the management of large amounts of stored data. The effectiveness of existing compression methods can vary greatly depending on the size of the source document. For example, an electronic image management (EIM) system configured to scan and store continuous tone images will have different image compression requirements than an application involving only text. The actual methods used to analyze the image compression requirements of the application and to select the best image compression method are complex, and it is clearly useful to present this guidance document to guide application and system development in selecting these methods. This guidance technical document is equivalent to ISO/TS12033:2001 "Guide to the Selection of Image Compression Methods for Electronic Imaging Documents" (English and Italian version). This guidance technical document is for reference only. Suggestions and opinions on this guidance technical document should be submitted to the State Council's standardization administrative department. This guidance technical document is proposed and coordinated by the National Document Imaging Technology Standardization Technical Committee (SAC/TC86). The guidance technical document is then reviewed by the National Document Imaging Technology Standardization Technical Committee. The main drafters of this guidance technical document are: Li Ming, Di Jianwei, Wang Kun. 1 Scope GB/Z19736-2005/IS0/TS12033:2001 Electronic Imaging ||Guide to Selecting a Compression Method for Document Images
This technical guidance document provides information for users or Electronic Image Management (EIM) system integrators to enable them to make decisions regarding the selection of a compression method for digital images of commercial documents. Its purpose is to provide information for analyzing the type of file and which compression method is most appropriate for a particular file in order to optimize the storage and use of the file. For users, this technical guidance document provides information about image compression methods available in hardware or software to assist the user in selecting a system with an embedded compression method. For equipment or software designers, it provides planning information. This technical guidance document is only applicable to static images in bit-mapped patterns. It also considers the selection of compression methods based on well-tested data. Compression algorithm for mathematical calculation.
2 Normative referenced documents
The clauses in the following documents become the clauses of this guiding technical document through reference in this guiding technical document. For any previous referenced documents, all subsequent amendments (excluding errata) or revisions are not applicable to this guiding technical document. However, all parties who submit documents for cooperation based on this guiding technical document are encouraged to study whether the latest versions of these documents can be used. For any undated referenced documents, their latest versions are applicable to this guiding technical document. ISO12651:1999 Vocabulary of electronic imaging
ITU-T Recommendation T.4:1999 Standardization of Category 3 fax terminals for document transmission ITU-T Recommendation T.6:1988 Category 4 fax equipment Alternative fax coding schemes and coding control functions 3 Terms and definitions
IS012651:1999 The following terms and definitions apply to this guidance document. 3.1
lossless compression
A compression algorithm that can recover all the original information of the compressed image. 3.2
lossy compression
A compression algorithm that loses some of the original information during compression, so that the decompressed image is only an approximation of the original image. Note: This type of algorithm is particularly useful for image compression if the fragments are imperceptible or almost imperceptible to the human eye and can be included. In this case, the compression ratio can be greatly improved. 3.3
resolution
The number of pixels per unit length.
dots per inch
The number of dots that a scanner (printer) can scan (print) per inch in the horizontal and vertical directions. GB/Z19736—2005/IS0/TS 12033:20013.5
Brightness
The visual experience that enables the observer to perceive millimeter. 3.6
Contrast
The difference between the highest density and the lowest density of an image. 3.7
Bit level
The number of bits used to define a pixel,
luminance
The luminous flux emitted by a surface,
chrominance
The color part of the video signal that includes hue and saturation but not brightness. Note: Low chroma means that the color picture appears white. 3.10
ITU-T Group 3 and Group 4ITU-T Group 3 and Group 4Two standard derivation algorithms proposed by ITEJ-T.
Joint Photographie Experts GroupJPEG
The popular name of the IS0/IEC10994 standard. 3.12
YKAONIKAca
Comite Consultatif International Telegraphique et Telephonique; International Telegraph and Telephone Consultative Committee
International Telecommunication Union
The former name of the Telecommunication Standardization Group (-T). 3.13
Compression ratiocompressianratlo
The ratio of the image file size before compression to the image file size after compression. 4 General
In document imaging systems.Users are concerned about the quality of archived images for two reasons: first, because it can affect the medium- and even long-term prospects of the imaging system; second, because they must choose imaging tools based on a constantly evolving technology. The digitization process inherently transforms information-rich images into immaterial images, thus changing the observer's perception of the image (picture). The observer may perceive the image as improved, but more often, the image quality is degraded. Fact: Images undergo a series of transformations at different stages of the digitalization process. At each stage, people try to keep the readability of the image within acceptable limits, but also limit the size of the image within the economic limits currently accepted. Among the stages of digitization, there is a stage of compression, whose specific purpose is to reduce the size of the image file. Some compression methods are reversible, that is, the decompression algorithm can restore the original digital information. These methods are lossless, and when the human eye sees the image, the image quality is not damaged. Other methods are lossy and will cause the image quality that the human eye can perceive to be reduced. By adjusting the parameters, the user can find a lossy compression method within acceptable limits. GB/Z 19736—2005/ES0/TS 12033;2001Although a large number of compression methods have been described in the technical literature, only a few have been stabilized according to industry standards. These are based on a limited number of principles: dominance of certain patterns, repetitiveness of patterns, and significant mathematical properties. In any one method, the number of parameters that can be modified by the user is relatively small. The choice of compression method and compression parameters is largely determined by the characteristics of the output file. Obviously, the graphic content of the file plays a key role in determining the method and its parameters. However, other factors that affect the characteristics of the application are also important (see Figure 1). The graphic content of the file itself is important for the digitization process. Thus, a photograph cannot be digitized in the same way if it is grayscale or based on a "pseudo-grayscale" method. The former will use JPEG compression, while the latter will need to use ITU or IBIG compression.
Therefore, before discussing compression methods, we need to review the types of files and how they are represented after digitization, see Figure 1.
Document usage
Digitalization
Technical features
Text and graphic content
Compression method
Text quality
Product presentation mode
Observation conditions
Figure 1 Interaction of various factors and compression methods 5 File types and digitization parameters
5.1 Overview
A file is a group of organized information intended to be presented to a user. Files can be single-page or multi-page and can contain any type of content, such as character content, graphic content, and various types of image content. The following file content can be found in various types of files. The following classification is somewhat arbitrary, and people can use these differences to understand how to handle a given file for a given application. 5.2 Types of files
Only the files most likely to be archived electronically (often called "word processing files") are given here. These files include: black text on a white background, or less commonly, color text on a color background; single-color or black-and-white prints;
mixed black-and-white or color files containing both text and photos reproduced by printing. 5.3 Classification and digitization of files
In order to determine the compression scheme, files can be described in the following five ways. For each type of file, the digitization method is briefly described.
GB/Z19736—2005/IS0/TS12033,20015. 3.2 Black and white documents
IYKAoNiKAca-
After digitizing pages of black and white printing (mainly text), binary images are generated, each pixel is reproduced using bits. This form of reproduction can also be applied to text files with colored backgrounds or characters, as well as line drawings. The most important digitization parameter is the resolution. The resolution must be determined according to the needs of visual perception and is limited by the entire imaging system (for example, 200 dpi for literary processing files and 300 dpi for digitized books). There are other factors involved in image processing, which vary with the type of image. For example, if we know that the image to be digitized is text, we will strive to generate black characters with sharp edges on a colored background. Therefore, we have the two parameters of brightness (adjusting the color of a pixel relative to a threshold) and contrast (adjusting the color of a pixel relative to the colors of surrounding pixels). 5. 3. 3 Grayscale Files
This form of reproduction is suitable for photographic files printed on photographic paper from black and white film. Digitization converts an originally continuous-tone file into a matrix of pixels encoding the blackness of a pixel in varying degrees. Thus, an 8-bit encoding yields 256 shades of gray. The number of shades or bit depth must be determined based on the needs of visual perception and the limitations of the entire imaging process. 5.3.4 Pseudo-grayscale Files
This type of file consists of an image that simulates gray using a variable distribution of black and white pixels. This can be in two cases: (1) the source file is a photographic reproduction of a text that was originally produced using printing technology and is itself a pseudo-grayscale file (the screen is formed using black pixels of varying sizes); (2) the source file is a real photograph that has been digitized in pseudo-grayscale form to reduce storage space or transmission time over the Internet (the "halftone" technique involves distributing varying numbers of black pixels in a fixed-size matrix). 5.3.5 Color Documents
This form of reproduction is suitable for photographic documents printed on photographic paper from color film. Another use is the mathematical color capture of business documents, where yellow highlights, colored borders, pencil lines, red pen lines, etc. are all part of the overall information capture. Color documents are intended to be restored in color, but can also be reproduced in grayscale. Color reproduction is based on the neurophysiological properties of the human eye, in particular the principle of "visual primary colors", that is, all colors can be reproduced by a combination of three primary colors. Thus, a color can be reproduced by three coordinates in a vector space based on the three primary colors, or by their linear combination.
The most commonly used color space uses red, green and blue. The colors are distinguished by the retinal cone cells in the eye. Another color space replaces these variables with one "brightness" variable and two "chrominance" variables. This color space is used to transmit television signals. In a digital color image, each pixel is reproduced by a combination of three components corresponding to the three primary colors. The bit depth used for a component determines the quality of the color: the standard of 8 bits per component can reproduce 256 (more than 16 million) different colors, and a total of 8 bits of reproduction transmitted by data communication networks is also quite common. 5. 3. 6 Mixed files
Many files intended for archiving consist of a number of text pages with graphic elements and/or photographic images. There is no completely satisfactory way to render such files:
—a binary reproduction would render the illustrations unreadable;
—a grayscale or color reproduction to preserve the illustrations would indeed provide the best quality, but would disproportionately increase the storage requirements in terms of the importance of the images (one must be aware that there may be a trade-off between resolution and bit depth for grayscale or color image files);
a halftone reproduction would reduce the readability of the characters. In mixed files, the text is considered more important, so a binary reproduction would be used, with black characters on a white background, and the photographs would either suffer or have to be separated from the text for a suitable reproduction. In most cases, the text and photographs can be automatically and successfully separated using a segmentation algorithm. Sometimes the segmentation results in the loss of information (e.g., when there is a caption under the photo, or an unusual print layout is used). 6 Compression Methods and Standards
6.1 RLE Compression (Run Length Encoding)
This method is accepted by most graphic image formats. It considers runs of the same symbol (or character in the case of ASCII text) in a data stream, each of which is encoded by the number of times the repeating element occurs and the length of the data stream. The RLE algorithm can operate at the bit, byte, or pixel level. The basic algorithm operates one line at a time, but some variants operate vertically, taking into account repeated symbols in adjacent lines. RLE is nominally lossless, but some variants discard low-order bits for efficiency, which results in loss. It is not very effective for text and complex photos, where long sequences are rare. It is most effective for images with large areas of uniform dithering. 6.2 I.Zw compression (Lempel-Ziv-Welch) A variation of this method can be used for some image formats. The algorithm does not search for repetitive elements in a fixed way, but rather uses a dictionary of previously received data streams that is continually filled as the algorithm processes an image (for this example, the dictionary is initialized with ASCII letters). 6.3 ITU-T Algorithms
6.3.1 Overview
ITU-T has defined a series of protocols for transmitting images by fax. Officially these protocols are called T.4 and T.6, however they are commonly referred to as G3 (Type 3) and G4 (Type 4). The compression methods used in archiving are variations of the ITU-T algorithms. The ITI-T algorithms include end-of-line and end-of-message codes to simplify the fax process. When these methods are used for archiving, these codes are redundant. ITU-T compression is based on a variation of the Huffman algorithm. ITI-T defines three fax standards that are used to compress binary images: - Type 3 Modified Huffman (MH); one-dimensional compression method (G3 1D); - Type 3 Modified Heavy Reduction (MR); - Type 4 Modified MR (MMR); two-dimensional compression method (G4), 6.3.2 type one-dimensional compression method (G3 1D) || tt || type three-dimensional compression method (G3 1T)), which is a variation of the Huffman algorithm. In a binary image, each scan line contains a number of variable length regions consisting of black or white pixels alternately. The Class 3 encoder determines the length of each black or white region (called a run length) and looks up the corresponding code in the Huffman matrix.
Compression is achieved because the codewords are shorter than the regions they represent. Each codeword represents the length of a region corresponding to white or black. The Class 3 algorithm is the basic compression method used in Class 3 faxes. The length of the codewords was determined when the method was created, based on statistical observations of printed and handwritten documents. The shortest codewords are assigned to run lengths with a high probability of occurrence. Note: Although the ITU-T compression method was originally designed for text files, it is also suitable for raster photos, but with lower efficiency. The order of the image index is represented by two types of codewords: configuration codewords and terminal codewords. Configuration codewords represent long regions, and terminal codewords represent short regions. Regions with a length between 0 and 63 bits are encoded in a terminal codeword. A bit stream between 64 and 2623 bits is encoded in a configuration codeword corresponding to the quotient of the length divided by 64, and a terminal codeword is added for the remainder. A bit stream longer than 2623 bits is encoded into a series of configuration codewords, and then a terminal codeword may be added.
This one-dimensional encoding scheme only eliminates a few redundant bits from left to right in each scan line, without reducing the redundant bits between upper and lower scan lines.
GB/Z19736—2005/IS0/TS12033:20016.3.33-type two-dimensional method (G32D) and 4-type method IYKAoNiKAca-Www.bzxZ.net
The 3-type one-dimensional method processes each line of the image separately, while the 3-type two-dimensional method takes advantage of the consistency between two consecutive lines in the same image, which are often very similar.
G32D is defined as an alternative to class 3, which is limited to a small number of rows between the embedded "-dimensional" rows. Class 4 uses the same algorithm.
Like G31D, the G32D algorithm uses turning points (\independent pixels\) that separate different colors in a single row. When creating the coded representation of the image, the algorithm considers not only the independent pixels in a single row, but also the independent pixels in two adjacent rows. Thus, in addition to the codewords used in G31D, the G32D and G4 methods also use codewords that represent the distance and relative distribution of independent pixels in two or more adjacent rows.
6. 4 JBIG Compression
JBIG is the abbreviation for Joint Bi-level Image Group. As the name suggests, this method is used for binary images. It is mainly used for text (this is the same as T. 4 and T. 6), but it can also be used for raster photos in printed documents (this is the same as T. 4 and T. 6). According to its creators, for plain text, JBIG is equivalent to T.4 and T.6 in performance; for raster images, it is 2 to 30 times more efficient. It is also lossless compression. This method uses incremental coding: it controls the size of the resolution. This coding system initially transmits the image at a low resolution (for example, 25dpi), and then the resolution is progressively doubled until the original resolution image is obtained. The advantages of the incremental coding method are twofold: a) it analyzes the image only to the necessary level of detail; b) it can modify the resolution according to the characteristics of the output peripheral or the perception of the observer (for example, once the image is recognized, the transmission can be terminated).
ISO/IEC11544 describes the JBIG method as a combination of program blocks. Most of these blocks have the dual purpose of saving code and speeding up processing. Code is saved when uniform color areas, repetitive areas and identical areas are detected between the low-resolution image and the high-resolution image.
6. 5JPEG compression
6. 5.1 Overview
In general, images are highly correlated from one pixel to the next. That is, if one pixel is a certain shade of gray, there is a good chance that its neighboring pixels are also similar shades of gray. This means that some redundant information can be discarded without seriously affecting the readability of the image. The JPEG standard uses this approach. JPEG covers a range of algorithms and two types of compression. One type is lossless compression, but it is not very efficient. The more common type is lossy compression, which involves several steps, some of which are lossy. At the heart of this type of compression is a mathematical transform called the discrete cosine transform (IXCT). 6.5.2 Discrete Cosine Transform (DCT)
The DCT is applied to each pixel in each block (a block of 8×8 pixels extracted from the digitized image). The DCT transforms the block into 64 frequency coefficients.
An interesting property of this transform is that it concentrates information in a small number of coefficients, while most of the other coefficients have values close to zero. 6.5.3 JPFG Steps
The JPEG method includes the following steps:
a) Discrete cosine transform;
b) Truncation caused by digital representation of data (discarding very low-order values);) "Quantization\transformation
d) "Post-encoding", that is, not only JPEG compression is used, but also Zeffman coding technology or algorithm compression. In addition to DCT, the most important step in IPEG is quantization. Re-encoding includes linear transformation of the coefficients obtained by DCT. This quantization is used to eliminate frequencies with small base values (mostly cheeks) and to roughly represent the remaining frequencies. GB/Z97362005/IS0/TS12033200
The quantization matrix controls both the obtained reduction ratio and the degradation of the image. It is possible to fix a "loss level". JFEG experts did not strictly define the quantization matrix in the standard. They only gave an example suitable for a 720×576 pixel TV screen. The quantization matrix example is based on a psychological and visual test of photos. 6.5.4JPEG Components
JPEG mainly describes the principles of compression. Some (like LCT) are specific to JPEG, while others (like Hofman coding) are not.
Although JPEG does not use a color space, it can take advantage of the human eye's greatest sensitivity to chrominance and brightness variables. Thus, in Y, Cr, Ch coding, it can subsample the chrominance information according to the brightness. JPEG also introduces aspects about the dynamic characteristics of image display, which are not directly related to the compression principle. These auxiliary aspects affect the visual perception of digitized images, making them key criteria for JPEG. - Interleaving: JPEG can transmit Y, Cr, Cb components in groups of 3, i.e. interleaved. - DCT sequential coding: the image is transmitted several times, and the quality is improved with each transmission. The first principle can be used with an interactive search system: most images are considered to be equal and are transmitted with the lowest quality, and only those images that are specifically searched are transmitted with the highest quality. 6.6 Fractal Compression
This method is based on fractal geometry, created by French mathematician Benoit Mandelbrot. Unlike analytic geometry, which uses infinitesimal elements composed of a network of lines to approximate complex curves, fractal geometry believes that dividing a curve (or surface) into smaller elements does not reduce its complexity. A special type of fractal is generated by the infinite repetition of the same pattern in the infinite subdivision elements of a given curve. Image compression using fractal methods is the reverse process of generating fractal images. The purpose of compression is not to generate an image from a given transformation, but to find a series of transformations that approximate a given digital image. Through the simulation of fractal geometry, the basic goal is to find the similarity between image elements obtained at different subdivision levels. If all small patterns can be considered similar to human elements, they do not need to be described in detail, thus achieving compression. In the fractal method, the algorithm attempts to compare image elements using short array transformations. The algorithm includes an image subdivision principle and a metric definition that allows determining how similar the pixels in the image are. 6.7 Wavelet Compression
This method is based on the work of Fourier in the 19th century. The Fourier series can represent any periodic function as a linear combination of sinusoidal functions of different order. The Fourier transform needed to be modified to accommodate discrete-valued functions such as those generated digitally, and to handle non-periodic functions. This gave rise to the DCT, which is the basis of JPEG. However, the Fourier transform is not suitable for sharp discontinuities in images. Wavelets form a series of non-sinusoidal mathematical functions. These functions can have peaks and can fall off rapidly, which means that these functions can be associated with non-periodic local features of the image. As in the LC1 method, wavelet compression involves replacing the digitally generated values with linear complex coefficients. The compression effect is achieved by the presence of coefficients close to zero, 7 Choice of Compression Parameters
7.1 Reasonable Compression
Since compression may or may not be lossy, our first inclination should be to use lossless compression methods. However, we are immediately faced with two problems:
a) Currently only ITUTG3 and G4 and JPEC have achieved standardization and large-scale industrialization: b) The first problem will inevitably lead us to prefer JPEG for photos, but the performance of lossless JPEG is much lower than that of lossy JPEG.
Therefore, in the case of IPEC, the most important thing is to decide which compression is more appropriate. To answer this question, we must seriously consider the role of the image in the imaging system. GB/Z19736—2005/ES0/TS12033:2001 There can be two extreme cases:
-YKAONKAca-
a) Either the source file has a high aesthetic value. The digitized image is required to have the highest possible quality; b) Or the content of the file is more important than its appearance, so as long as the digitized file is still readable, a certain compromise between quality and performance can be achieved.
The first case should generally exclude lossy compression-while the second case is to use compression as a means to reduce the size of the image file, even if there is a trade-off.
To some extent, the architecture of the imaging system may determine whether quality or performance is more important. Alternatively, one of the following decisions may be made:
Store uncompressed files for editing and copying, and distribute compressed files when they are needed for display on a computer screen (this is the source of the "image recording", as opposed to the actual illustrated file; extract a portion of the file. Digitize at above-average quality, that is, use lossless compression: · Store both paper and digitized files, and when retrieval is required, only those paper files that require high-quality reproduction are digitized.
7.2 Choice of compression method
Once the decision is made to compress the images, the user must choose the compression method that best suits the various types of files in the collection. Table 1 summarizes the possible choices.
Table 1 Various Compression method
Type of file
Line drawing
Photographic drawing
Line drawing
Photographic film
Photo on medium
(Simplified shapes and colors)
Photo on photographic medium
(Simplified shapes and colors)
Photo on photographic medium
(Real-world image)
Photo on photographic medium
(Real-world image)
7,3Adjustments for JPEG compression
Compression method
Typical size before compression
jno KB
Typical compression ratio
Typical size after compression
Compression ratio cannot be defined directly in JPEG. Moreover, such an adjustment would be meaningless, since in all compression methods the compression ratio varies depending on the characteristics of the image. However, IPEG does allow the user to modify the compression ratio indirectly by adjusting the quality level. Obviously, the quality level is inversely proportional to the compression ratio (the highest quality level has the lowest compression ratio). The definition of quality is not static. Each hardware or software product has its own specific definition. Quality levels are generally defined in the ratio of % to GB/Z19736—2005/S0/TS12033:2001100%. To adjust the compression ratio in JPE, the user modifies the quality level. , seek the lowest possible quality level while preserving the appearance of the original file.
For most samples, a 5% magic level is generally considered the best. However, users will have to find the best level for their own needs. Therefore, when determining the quality level, it is advisable to digitize a sample file that is archived and consult with a variety of potential viewers to see how they feel about the results. In JPEG, too high a compression ratio will cause the image to degrade, making the areas generated by the method appear to be flat and valves appear between areas.
This defect can be eliminated by increasing the resolution of the digitization, although this will increase the size of the image file. On the other hand, when the source image has shapes with contrasting colors, compression is likely to cause the circles to become blurred. In this case, the chrominance should be subsampled to the luminance. Thus, a compromise needs to be made between resolution and quality level. 8 Conclusion
The choice of image compression method depends on the evolving state of the art, as well as the rapidly changing industry and commercial standards. Obviously it is difficult to predict what major technological developments will occur in the next few years. However, we can expect certain trends to continue. In particular, the two trends in compression methods for text and photographic documents are likely to continue. ITU-TG3 and G4 and JPEG are the most frequently used methods at present, but people are looking forward to more powerful methods. Therefore, users will have to choose between lossless methods for text and lossy methods for photographic documents. For photographic documents, the algorithms used will be similar to those used in JPEG... with different mathematical functions. GB/Z19736—2005/IS0/TS12033.2001 尊考文档
YKAoNiKAca
Information Technology Digital Coding and Compression of Continuous-Tone Still Images: Requirements and Guidelines [1] IS0/1EC10928-1.1994
[2] ISO/1FC 10928-2:1995
Information technology - Digital coding and compression of continuous-tone still images: Conformance test [3] ISO/IEC 10994:1992 Information technology - 90 mm floppy disks with a density of 31831 flux turns/radian and 80 magnetic fields per side, modified frequency modulation, for use in data interchange ISO 303 type serial compression
[4] ISO/IEC 11544:1993 Information technology - Coding and reproduction of visual and audio information Moore's Imaging Dictionary: The Official Dictionary of Electronic Documents and Image Processing: New York: Library of Telecommunications, 1993 [5]7 Wavelet Compression
This method is based on the work of Fourier in the 19th century. The Fourier series can represent any periodic function as a linear combination of sinusoidal functions of different order. The Fourier transform needed to be modified to accommodate discrete-valued functions such as those generated digitally, and to handle non-periodic functions. This gave rise to the DCT, which is the basis of JPEG. However, the Fourier transform is not suitable for sharp discontinuities in images. Wavelets form a series of non-sinusoidal mathematical functions. These functions can have peaks and can fall off rapidly, which means that these functions can be associated with non-periodic local features of the image. As in the LC1 method, wavelet compression involves replacing the digitally generated values with linear complex coefficients. The compression effect is achieved by the presence of coefficients close to zero, 7 Choice of Compression Parameters
7.1 Reasonable Compression
Since compression may or may not be lossy, our first inclination should be to use lossless compression methods. However, we are immediately faced with two problems:
a) Currently only ITUTG3 and G4 and JPEC have achieved standardization and large-scale industrialization: b) The first problem will inevitably lead us to prefer JPEG for photos, but the performance of lossless JPEG is much lower than that of lossy JPEG.
Therefore, in the case of IPEC, the most important thing is to decide which compression is more appropriate. To answer this question, we must seriously consider the role of the image in the imaging system. GB/Z19736—2005/ES0/TS12033:2001 There can be two extreme cases:
-YKAONKAca-
a) Either the source file has a high aesthetic value. The digitized image is required to have the highest possible quality; b) Or the content of the file is more important than its appearance, so as long as the digitized file is still readable, a certain compromise between quality and performance can be achieved.
The first case should generally exclude lossy compression-while the second case is to use compression as a means to reduce the size of the image file, even if there is a trade-off.
To some extent, the architecture of the imaging system may determine whether quality or performance is more important. Alternatively, one of the following decisions may be made:
Store uncompressed files for editing and copying, and distribute compressed files when they are needed for display on a computer screen (this is the source of the "image recording", as opposed to the actual illustrated file; extract a portion of the file. Digitize at above-average quality, that is, use lossless compression: · Store both paper and digitized files, and when retrieval is required, only those paper files that require high-quality reproduction are digitized.
7.2 Choice of compression method
Once the decision is made to compress the images, the user must choose the compression method that best suits the various types of files in the collection. Table 1 summarizes the possible choices.
Table 1 Various Compression method
Type of file
Line drawing
Photographic drawing
Line drawing
Photographic film
Photo on medium
(Simplified shapes and colors)
Photo on photographic medium
(Simplified shapes and colors)
Photo on photographic medium
(Real-world image)
Photo on photographic medium
(Real-world image)
7,3Adjustments for JPEG compression
Compression method
Typical size before compression
jno KB
Typical compression ratio
Typical size after compression
Compression ratio cannot be defined directly in JPEG. Moreover, such an adjustment would be meaningless, since in all compression methods the compression ratio varies depending on the characteristics of the image. However, IPEG does allow the user to modify the compression ratio indirectly by adjusting the quality level. Obviously, the quality level is inversely proportional to the compression ratio (the highest quality level has the lowest compression ratio). The definition of quality is not static. Each hardware or software product has its own specific definition. Quality levels are generally defined in the ratio of % to GB/Z19736—2005/S0/TS12033:2001100%. To adjust the compression ratio in JPE, the user modifies the quality level. , seek the lowest possible quality level while preserving the appearance of the original file.
For most samples, a 5% magic level is generally considered the best. However, users will have to find the best level for their own needs. Therefore, when determining the quality level, it is advisable to digitize a sample file that is archived and consult with a variety of potential viewers to see how they feel about the results. In JPEG, too high a compression ratio will cause the image to degrade, making the areas generated by the method appear to be flat and valves appear between areas.
This defect can be eliminated by increasing the resolution of the digitization, although this will increase the size of the image file. On the other hand, when the source image has shapes with contrasting colors, compression is likely to cause the circles to become blurred. In this case, the chrominance should be subsampled to the luminance. Thus, a compromise needs to be made between resolution and quality level. 8 Conclusion
The choice of image compression method depends on the evolving state of the art, as well as the rapidly changing industry and commercial standards. Obviously it is difficult to predict what major technological developments will occur in the next few years. However, we can expect certain trends to continue. In particular, the two trends in compression methods for text and photographic documents are likely to continue. ITU-TG3 and G4 and JPEG are the most frequently used methods at present, but people are looking forward to more powerful methods. Therefore, users will have to choose between lossless methods for text and lossy methods for photographic documents. For photographic documents, the algorithms used will be similar to those used in JPEG... with different mathematical functions. GB/Z19736—2005/IS0/TS12033.2001 尊考文档
YKAoNiKAca
Information Technology Digital Coding and Compression of Continuous-Tone Still Images: Requirements and Guidelines [1] IS0/1EC10928-1.1994
[2] ISO/1FC 10928-2:1995
Information technology - Digital coding and compression of continuous-tone still images: Conformance test [3] ISO/IEC 10994:1992 Information technology - 90 mm floppy disks with a density of 31831 flux turns/radian and 80 magnetic fields per side, modified frequency modulation, for use in data interchange ISO 303 type serial compression
[4] ISO/IEC 11544:1993 Information technology - Coding and reproduction of visual and audio information Moore's Imaging Dictionary: The Official Dictionary of Electronic Documents and Image Processing: New York: Library of Telecommunications, 1993 [5]7 Wavelet Compression
This method is based on the work of Fourier in the 19th century. The Fourier series can represent any periodic function as a linear combination of sinusoidal functions of different order. The Fourier transform needed to be modified to accommodate discrete-valued functions such as those generated digitally, and to handle non-periodic functions. This gave rise to the DCT, which is the basis of JPEG. However, the Fourier transform is not suitable for sharp discontinuities in images. Wavelets form a series of non-sinusoidal mathematical functions. These functions can have peaks and can fall off rapidly, which means that these functions can be associated with non-periodic local features of the image. As in the LC1 method, wavelet compression involves replacing the digitally generated values with linear complex coefficients. The compression effect is achieved by the presence of coefficients close to zero, 7 Choice of Compression Parameters
7.1 Reasonable Compression
Since compression may or may not be lossy, our first inclination should be to use lossless compression methods. However, we are immediately faced with two problems:
a) Currently only ITUTG3 and G4 and JPEC have achieved standardization and large-scale industrialization: b) The first problem will inevitably lead us to prefer JPEG for photos, but the performance of lossless JPEG is much lower than that of lossy JPEG.
Therefore, in the case of IPEC, the most important thing is to decide which compression is more appropriate. To answer this question, we must seriously consider the role of the image in the imaging system. GB/Z19736—2005/ES0/TS12033:2001 There can be two extreme cases:
-YKAONKAca-
a) Either the source file has a high aesthetic value. The digitized image is required to have the highest possible quality; b) Or the content of the file is more important than its appearance, so as long as the digitized file is still readable, a certain compromise between quality and performance can be achieved.
The first case should generally exclude lossy compression-while the second case is to use compression as a means to reduce the size of the image file, even if there is a trade-off.
To some extent, the architecture of the imaging system may determine whether quality or performance is more important. Alternatively, one of the following decisions may be made:
Store uncompressed files for editing and copying, and distribute compressed files when they are needed for display on a computer screen (this is the source of the "image recording", as opposed to the actual illustrated file; extract a portion of the file. Digitize at above-average quality, that is, use lossless compression: · Store both paper and digitized files, and when retrieval is required, only those paper files that require high-quality reproduction are digitized.
7.2 Choice of compression method
Once the decision is made to compress the images, the user must choose the compression method that best suits the various types of files in the collection. Table 1 summarizes the possible choices.
Table 1 Various Compression method
Type of file
Line drawing
Photographic drawing
Line drawing
Photographic film
Photo on medium
(Simplified shapes and colors)
Photo on photographic medium
(Simplified shapes and colors)
Photo on photographic medium
(Real-world image)
Photo on photographic medium
(Real-world image)
7,3Adjustments for JPEG compression
Compression method
Typical size before compression
jno KB
Typical compression ratio
Typical size after compression
Compression ratio cannot be defined directly in JPEG. Moreover, such an adjustment would be meaningless, since in all compression methods the compression ratio varies depending on the characteristics of the image. However, IPEG does allow the user to modify the compression ratio indirectly by adjusting the quality level. Obviously, the quality level is inversely proportional to the compression ratio (the highest quality level has the lowest compression ratio). The definition of quality is not static. Each hardware or software product has its own specific definition. Quality levels are generally defined in the ratio of % to GB/Z19736—2005/S0/TS12033:2001100%. To adjust the compression ratio in JPE, the user modifies the quality level. , seek the lowest possible quality level while preserving the appearance of the original file.
For most samples, a 5% magic level is generally considered the best. However, users will have to find the best level for their own needs. Therefore, when determining the quality level, it is advisable to digitize a sample file that is archived and consult with a variety of potential viewers to see how they feel about the results. In JPEG, too high a compression ratio will cause the image to degrade, making the areas generated by the method appear to be flat and valves appear between areas.
This defect can be eliminated by increasing the resolution of the digitization, although this will increase the size of the image file. On the other hand, when the source image has shapes with contrasting colors, compression is likely to cause the circles to become blurred. In this case, the chrominance should be subsampled to the luminance. Thus, a compromise needs to be made between resolution and quality level. 8 Conclusion
The choice of image compression method depends on the evolving state of the art, as well as the rapidly changing industry and commercial standards. Obviously it is difficult to predict what major technological developments will occur in the next few years. However, we can expect certain trends to continue. In particular, the two trends in compression methods for text and photographic documents are likely to continue. ITU-TG3 and G4 and JPEG are the most frequently used methods at present, but people are looking forward to more powerful methods. Therefore, users will have to choose between lossless methods for text and lossy methods for photographic documents. For photographic documents, the algorithms used will be similar to those used in JPEG... with different mathematical functions. GB/Z19736—2005/IS0/TS12033.2001 尊考文档
YKAoNiKAca
Information Technology Digital Coding and Compression of Continuous-Tone Still Images: Requirements and Guidelines [1] IS0/1EC10928-1.1994
[2] ISO/1FC 10928-2:1995
Information technology - Digital coding and compression of continuous-tone still images: Conformance test [3] ISO/IEC 10994:1992 Information technology - 90 mm floppy disks with a density of 31831 flux turns/radian and 80 magnetic fields per side, modified frequency modulation, for use in data interchange ISO 303 type serial compression
[4] ISO/IEC 11544:1993 Information technology - Coding and reproduction of visual and audio information Moore's Imaging Dictionary: The Official Dictionary of Electronic Documents and Image Processing: New York: Library of Telecommunications, 1993 [5]so that some lossless compression can be achieved.
The first case should generally rule out lossy compression - while the second case, where compression as a means of reducing the image file size is permissible even with lossy compression.
To some extent, the architecture of the imaging system can determine whether quality or performance is more important. Alternatively, one of the following decisions may be made:
Store uncompressed files for editing and copying, and when they are needed for display on a computer screen, distribute compressed files (this is the source of the "image recording", as opposed to the file of the actual illustration); extract a portion of the file. Digitize at above-average quality, i.e., use lossless compression: · Store both paper and digitized files, and when retrieval is required, only those paper files that require high-quality reproduction are digitized.
7.2 Choice of compression method
Once the decision is made to compress the images, the user must choose the compression method that best suits the various types of files in the collection. Table 1 summarizes the possible choices.
Table 1 Various Compression method
Type of file
Line drawing
Photographic drawing
Line drawing
Photographic film
Photo on medium
(Simplified shapes and colors)
Photo on photographic medium
(Simplified shapes and colors)
Photo on photographic medium
(Real-world image)
Photo on photographic medium
(Real-world image)
7,3Adjustments for JPEG compression
Compression method
Typical size before compression
jno KB
Typical compression ratio
Typical size after compression
Compression ratio cannot be defined directly in JPEG. Moreover, such an adjustment would be meaningless, since in all compression methods the compression ratio varies depending on the characteristics of the image. However, IPEG does allow the user to modify the compression ratio indirectly by adjusting the quality level. Obviously, the quality level is inversely proportional to the compression ratio (the highest quality level has the lowest compression ratio). The definition of quality is not static. Each hardware or software product has its own specific definition. Quality levels are generally defined in the ratio of % to GB/Z19736—2005/S0/TS12033:2001100%. To adjust the compression ratio in JPE, the user modifies the quality level. , seek the lowest possible quality level while preserving the appearance of the original file.
For most samples, a 5% magic level is generally considered the best. However, users will have to find the best level for their own needs. Therefore, when determining the quality level, it is advisable to digitize a sample file that is archived and consult with a variety of potential viewers to see how they feel about the results. In JPEG, too high a compression ratio will cause the image to degrade, making the areas generated by the method appear to be flat and valves appear between areas.
This defect can be eliminated by increasing the resolution of the digitization, although this will increase the size of the image file. On the other hand, when the source image has shapes with contrasting colors, compression is likely to cause the circles to become blurred. In this case, the chrominance should be subsampled to the luminance. Thus, a compromise needs to be made between resolution and quality level. 8 Conclusion
The choice of image compression method depends on the evolving state of the art, as well as the rapidly changing industry and commercial standards. Obviously it is difficult to predict what major technological developments will occur in the next few years. However, we can expect certain trends to continue. In particular, the two trends in compression methods for text and photographic documents are likely to continue. ITU-TG3 and G4 and JPEG are the most frequently used methods at present, but people are looking forward to more powerful methods. Therefore, users will have to choose between lossless methods for text and lossy methods for photographic documents. For photographic documents, the algorithms used will be similar to those used in JPEG... with different mathematical functions. GB/Z19736—2005/IS0/TS12033.2001 尊考文档
YKAoNiKAca
Information Technology Digital Coding and Compression of Continuous-Tone Still Images: Requirements and Guidelines [1] IS0/1EC10928-1.1994
[2] ISO/1FC 10928-2:1995
Information technology - Digital coding and compression of continuous-tone still images: Conformance test [3] ISO/IEC 10994:1992 Information technology - 90 mm floppy disks with a density of 31831 flux turns/radian and 80 magnetic fields per side, modified frequency modulation, for use in data interchange ISO 303 type serial compression
[4] ISO/IEC 11544:1993 Information technology - Coding and reproduction of visual and audio information Moore's Imaging Dictionary: The Official Dictionary of Electronic Documents and Image Processing: New York: Library of Telecommunications, 1993 [5]so that some lossless compression can be achieved.
The first case should generally rule out lossy compression - while the second case, where compression as a means of reducing the image file size is permissible even with lossy compression.
To some extent, the architecture of the imaging system can determine whether quality or performance is more important. Alternatively, one of the following decisions may be made:
Store uncompressed files for editing and copying, and when they are needed for display on a computer screen, distribute compressed files (this is the source of the "image recording", as opposed to the file of the actual illustration); extract a portion of the file. Digitize at above-average quality, i.e., use lossless compression: · Store both paper and digitized files, and when retrieval is required, only those paper files that require high-quality reproduction are digitized.
7.2 Choice of compression method
Once the decision is made to compress the images, the user must choose the compression method that best suits the various types of files in the collection. Table 1 summarizes the possible choices.
Table 1 Various Compression method
Type of file
Line drawing
Photographic drawing
Line drawing
Photographic film
Photo on medium
(Simplified shapes and colors)
Photo on photographic medium
(Simplified shapes and colors)
Photo on photographic medium
(Real-world image)
Photo on photographic medium
(Real-world image)
7,3Adjustments for JPEG compression
Compression method
Typical size before compression
jno KB
Typical compression ratio
Typical size after compression
Compression ratio cannot be defined directly in JPEG. Moreover, such an adjustment would be meaningless, since in all compression methods the compression ratio varies depending on the characteristics of the image. However, IPEG does allow the user to modify the compression ratio indirectly by adjusting the quality level. Obviously, the quality level is inversely proportional to the compression ratio (the highest quality level has the lowest compression ratio). The definition of quality is not static. Each hardware or software product has its own specific definition. Quality levels are generally defined in the ratio of % to GB/Z19736—2005/S0/TS12033:2001100%. To adjust the compression ratio in JPE, the user modifies the quality level. , seek the lowest possible quality level while preserving the appearance of the original file.
For most samples, a 5% magic level is generally considered the best. However, users will have to find the best level for their own needs. Therefore, when determining the quality level, it is advisable to digitize a sample file that is archived and consult with a variety of potential viewers to see how they feel about the results. In JPEG, too high a compression ratio will cause the image to degrade, making the areas generated by the method appear to be flat and valves appear between areas.
This defect can be eliminated by increasing the resolution of the digitization, although this will increase the size of the image file. On the other hand, when the source image has shapes with contrasting colors, compression is likely to cause the circles to become blurred. In this case, the chrominance should be subsampled to the luminance. Thus, a compromise needs to be made between resolution and quality level. 8 Conclusion
The choice of image compression method depends on the evolving state of the art, as well as the rapidly changing industry and commercial standards. Obviously it is difficult to predict what major technological developments will occur in the next few years. However, we can expect certain trends to continue. In particular, the two trends in compression methods for text and photographic documents are likely to continue. ITU-TG3 and G4 and JPEG are the most frequently used methods at present, but people are looking forward to more powerful methods. Therefore, users will have to choose between lossless methods for text and lossy methods for photographic documents. For photographic documents, the algorithms used will be similar to those used in JPEG... with different mathematical functions. GB/Z19736—2005/IS0/TS12033.2001 尊考文档
YKAoNiKAca
Information Technology Digital Coding and Compression of Continuous-Tone Still Images: Requirements and Guidelines [1] IS0/1EC10928-1.1994
[2] ISO/1FC 10928-2:1995
Information technology - Digital coding and compression of continuous-tone still images: Conformance test [3] ISO/IEC 10994:1992 Information technology - 90 mm floppy disks with a density of 31831 flux turns/radian and 80 magnetic fields per side, modified frequency modulation, for use in data interchange ISO 303 type serial compression
[4] ISO/IEC 11544:1993 Information technology - Coding and reproduction of visual and audio information Moore's Imaging Dictionary: The Official Dictionary of Electronic Documents and Image Processing: New York: Library of Telecommunications, 1993 [5]Too high a compression ratio will result in a degradation of the image quality, making the areas generated by the method almost monochromatic and valving between areas. This defect can be eliminated by increasing the resolution of the digitization, although this will increase the size of the image file. On the other hand, when the source image has shapes with strong contrasting colors, compression is likely to cause blurring. In this case, the chroma should be subsampled to the brightness. Thus, a compromise needs to be made between resolution and quality. 8 Conclusion The choice of image compression method depends on the evolving state of the art, as well as on the rapidly changing industrial and commercial standards. Obviously, it is difficult to predict what major technological developments will occur in the next few years. However, we can expect certain continuing trends. In particular, the two trends in compression methods for text and photographic files are likely to continue. ITU-TG3 and G4 and JPEG are the most frequently used methods at present, but people expect more powerful methods to appear. Therefore, users will have to choose between lossless methods for text and lossy methods for photographic files. For photographic files, the algorithm used will be similar to the algorithm used in JPEG... It has different mathematical functions. GB/Z19736-2005/IS0/TS12033.2001 References
YKAoNiKAca
Information Technology Digital Coding and Compression of Continuous-Tone Still Images: Requirements and Guidelines [1] IS0/1EC10928-1.1994
[2] ISO/1FC 10928-2:1995
Information technology - Digital coding and compression of continuous-tone still images: Conformance test [3] ISO/IEC 10994:1992 Information technology - 90 mm floppy disks with a density of 31831 flux turns/radian and 80 magnetic fields per side, modified frequency modulation, for use in data interchange ISO 303 type serial compression
[4] ISO/IEC 11544:1993 Information technology - Coding and reproduction of visual and audio information Moore's Imaging Dictionary: The Official Dictionary of Electronic Documents and Image Processing: New York: Library of Telecommunications, 1993 [5]Too high a compression ratio will result in a degradation of the image quality, making the areas generated by the method almost monochromatic and valving between areas. This defect can be eliminated by increasing the resolution of the digitization, although this will increase the size of the image file. On the other hand, when the source image has shapes with strong contrasting colors, compression is likely to cause blurring. In this case, the chroma should be subsampled to the brightness. Thus, a compromise needs to be made between resolution and quality. 8 Conclusion The choice of image compression method depends on the evolving state of the art, as well as on the rapidly changing industrial and commercial standards. Obviously, it is difficult to predict what major technological developments will occur in the next few years. However, we can expect certain continuing trends. In particular, the two trends in compression methods for text and photographic files are likely to continue. ITU-TG3 and G4 and JPEG are the most frequently used methods at present, but people expect more powerful methods to appear. Therefore, users will have to choose between lossless methods for text and lossy methods for photographic files. For photographic files, the algorithm used will be similar to the algorithm used in JPEG... It has different mathematical functions. GB/Z19736-2005/IS0/TS12033.2001 References
YKAoNiKAca
Information Technology Digital Coding and Compression of Continuous-Tone Still Images: Requirements and Guidelines [1] IS0/1EC10928-1.1994
[2] ISO/1FC 10928-2:1995
Information technology - Digital coding and compression of continuous-tone still images: Conformance test [3] ISO/IEC 10994:1992 Information technology - 90 mm floppy disks with a density of 31831 flux turns/radian and 80 magnetic fields per side, modified frequency modulation, for use in data interchange ISO 303 type serial compression
[4] ISO/IEC 11544:1993 Information technology - Coding and reproduction of visual and audio information Moore's Imaging Dictionary: The Official Dictionary of Electronic Documents and Image Processing: New York: Library of Telecommunications, 1993 [5]
Tip: This standard content only shows part of the intercepted content of the complete standard. If you need the complete standard, please go to the top to download the complete standard document for free.