Text preview for : Digital Processes.pdf part of Digital Processes
Back to : Digital Processes.pdf | Home
Digital Processes
Digital Scanning
Basic concepts
Analog Machines
Example: Model A219
Digital Scanning Image Processing Printing
An exposure lamp illuminates the original. Mirrors reflect light from the original directly onto the
a219d507.wmf
photoconductor. This light writes a latent image on the photoconductor. This image is then developed with toner and transferred to the copy paper.
1 April 2001 8-Jul-99 Page 186
Digital Processes
Digital Scanning
Digital Machines
Example: Model A193
a193v505.wmf
The big difference with scanners in digital machines is that the light reflected from the original does not pass directly to the photoconductor. The light is reflected onto a light-sensitive element, such as a CCD (Charge Coupled Device). This device converts the light into an analog electrical signal. Circuits inside the machine convert this signal into a digital signal. This signal then passes to a laser diode, which emits a laser beam to write a latent image on the photoconductor. So, in a digital machine, there is a lot of electronics between the light reflected off the original and the light arriving at the photoconductor.
1 April 2001
Page 187
Digital Processes
Digital Scanning
Digital Signals
Digital signals consist of binary code. When scanning an original, binary code is used to represent the brightness of each pixel of the image. In the most simple of systems, there are only two values for each pixel: 0 and 1, for black and white. However, most machines use 4 or 8 bits. In a four-bit system, there are 16 possible values for each pixel. This allows black, white, and 14 shades of grey in between. Similarly, in an eight-bit system, there are 256 possible values for each pixel. This allows black, white, and 254 shades of grey in between (see the diagram).
digdata.wmf
1 April 2001
Page 188
Digital Processes
Digital Scanning
Digital Images
Overview
Analog machines transfer an optical image of the original directly onto the photoconductor. Digital machines break the image up into small dots, known as picture elements, or pixels for short. The example shows the image that the machine builds up of a fax machine test chart. This may seem to be a rather inaccurate representation. However, digital signals can be manipulated to enhance the image and create special effects. Also, digital images can be used immediately, or stored for later use (see Image Files). The size of the pixels (smaller pixels yield greater `resolution') depends on several factors related to the scanner and printer hardware. (The software may also be set up to alter the resolution in various ways, but we shall look at hardware in this section.)
1 April 2001
testchrt.wmf
Page 189
Digital Processes
Digital Scanning
Scanner Resolution
There are two points to consider: the image detector (typically a CCD) and the scanner motor
CCD
The CCD (charge-coupled device) is a line of photosensitive elements. The output of the CCD represents one line across the page. Each element of the CCD generates one picture element of the line. So the CCD resolution is the resolution of the scanner across the page (this is also known as the `main scan'). The more elements there are per unit length, the finer the resolution. Typical CCDs have 200 or 400 elements per inch (or, for Group 3 fax machines operating in metric units, 8 or 16 elements per mm).
CCD
Elements
ccdpixel.wmf
1 April 2001
Page 190
Digital Processes
Digital Scanning
Scanner or ADF Motor
Example: Model A229, ADF mode The scanner or ADF motor is normally a stepper motor. The distance fed by each step of the motor determines the resolution of the scan down the page (also known as the `sub scan' direction). Typical resolutions are 200, 300, or 400 lines per inch (or for Group 3 fax machines, 3.85, 7.7, or 15.4 lines per mm). To scan an image, the CCD scans a line. Then the scanner motor feeds the page one line, and the CCD scans another line. This is repeated until the entire page has been scanned.
Main scan Sub scan
a229d651.wmf
1 April 2001
Page 191
Digital Processes
Digital Scanning
Scanner Output
Each element of the CCD generates a voltage which represents the intensity of the light reflected onto it from the document. The signals from all the elements are output in sequence, to generate an analog signal that represents the line that is currently being scanned. The upper diagram on the right shows an example of output from a line on a page which is all white except for a black shape on the left of the page. After the line has been scanned, the scanner moves the document forward one scan line width to move the next scan line into position. Then, the CCD reads the next scan line. The bottom diagram shows the next line being scanned.
scanlin1
CCD Output W hite Black
CCD
SCAN LINE
W hite CCD Output Black
CCD
SCAN LINE
scanlin2.wmf
1 April 2001
Page 192
Digital Processes
Digital Scanning
The signals from each consecutive scan line are strung together end to end, and sent out as an analog signal. The diagram opposite shows what the video signal would be like for the two consecutive scan lines shown in the previous two diagrams. The output is then processed as described in Image Processing.
VIDEO SIGNAL One scan line White Black One scan line
Etc
ccdsig.wmf
The next few pages show the basics about how the processed data is printed.
1 April 2001
Page 193
Digital Processes
Digital Scanning
Printer Resolution
The output from the scanner is converted to a laser diode drive signal. The laser beam then writes a latent image of the original on the photoconductor. There are two points to consider: the laser beam as it arrives on the photoconductor, and the speed of the photoconductor. Example: Model H006, using a master belt Exposure of the photoconductor to the laser beam creates the latent image. To make the main scan, the laser beam moves across the photoconductor. The resolution depends on the speed of the laser beam's motion across the photoconductor and on the frequency of the laser beam on/off switching clock. To make the sub scan, the photoconductor rotates. The resolution depends on the speed that the photoconductor rotates. In multifunctional machines, laser engines have to be able to print at a range of resolutions: 400 dpi for copying and Group 4 fax, 600 dpi for printing, and 16 x 15.4 dots per mm (391.2 x 406.4 dpi) for Group 3 fax.
Sub Scan (Photoconductor Rotation)
Main Scan (Laser Beam Motion)
laserprt.wmf
For full details of the laser optic system, see the Laser Printing section.
1 April 2001 Page 194
Digital Processes
Digital Scanning
The cross section of the beam on the master (i.e., the size of each printed dot) varies from model to model; it is roughly circular. In the example shown, from a Group 3 fax machine, the diameter is about 80 µm. This means that the printed dots overlap each other slightly, as shown in the diagram. 80 µm is about 12 dots per mm, and 90 µm is about 11 dots per mm.
laserdot.wmf
However, the printer resolution is 16 x 15.4 dots per mm for a Group 3 fax machine. The dots are larger than this resolution, so they overlap. This results in a better image than if there were no overlap. Generally, the laser beam switches off between pixels, even between black pixels. Note that, unlike the scanner/ADF motors, the motor that drives the photoconductor is normally a dc motor, not a stepper motor. Therefore, in theory, the main scan lines written across the photoconductor will be sloping very slightly. For more details, see the Laser Printing section.
1 April 2001
Page 195
Digital Processes
Digital Scanning
Printer Output
During the copy cycle, the photoconductor is charged to about 900 V (see Photocopying Processes Charge). The laser beam writes a latent image on the photoconductor. The charge on irradiated areas drops significantly, typically to between 0 and -100 V. (Voltage values differ from model to model.) The area of the photoconductor that is irradiated depends on whether the 'write to white' or 'write to black' method is being used.
ORIGINAL WRITE TO WHITE WRITE TO BLACK
Irradiated Areas
laserwrt.wmf
1 April 2001
Page 196
Digital Processes
Image Processing
Image Processing
Introduction
This section describes how digital machines convert the image from a scanned original into digital data. This section also describes techniques for processing the digital data, so that the printout is as close to the original as possible. For example, techniques used to process a business letter will be different from those used to process an original containing photographs. Each model implements these techniques in different ways, and some models do not implement all the techniques. In addition, the order of steps may be slightly different from that presented here. This section will provide a general description, with examples from various models. The techniques used by black-and-white machines and color machines are different. Also, blackand-white machines can use two different types of image sensor in the scanner. As a result, this section will be divided into three sub-sections, as follows. · Black and White Machines - CCD Systems This section describes black-and-white models that use a CCD (Charge Coupled Device). This is the standard method for mainstream digital machines. · Black and White Machines - CIS Systems This section describes black-and-white models that use a CIS (Contact Image Sensor). This type of system is often used in lower-priced models. · Color Machines This section describes image processing for color machines. These use a CCD of a different type, to generate data for the three primary colors.
1 April 2001 Page 197
Digital Processes
Image Processing
Black and White CCD Systems
Overview
CCD
SBU
Memory Control ICs
GA1 LD Driver Drum LD Driver
LD Controller (GAVD)
HDD
IPU GA2
LDDR
SBICU
a229d578 wmf
The diagram shows a typical example of an image processing circuit. An exposure lamp illuminates the original. Light reflected from the original is reflected through a lens to the CCD.
1 April 2001
Page 198
Digital Processes
Image Processing
The CCD generates an analog signal from the light. The voltage of the signal varies with the intensity of the light. The CCD is mounted on a board called the SBU (Sensor Board Unit). The analog output from the CCD must be converted to a digital signal. In the above example, the analog-to-digital conversion circuits are on the SBU board. The digital signal is then processed, using large-scale integrated circuits, like the IPU (Image Processing Unit) in the above example. Some of the processes require enough working memory to store a page of image data. The data may then be stored temporarily on a hard disk until it is time for printing. The data then passes to the laser diode controller and laser diode driver. After data processing, each pixel scanned from the original is represented by a number of bits (eight is a typical number), or only one bit (0: White, 1: Black), depending on the type of digital processing used. Also, the image may be enlarged or reduced. In this case, pixels will be deleted or artificially created to make the new image.
Scanner Lamps and the Shading Plate
Fluorescent lamp: The ends of the lamp are not so bright as the center. To compensate for this, the light reflected from the original goes through a shading plate before it reaches the CCD. The shading plate allows more light to pass through from the ends of the lamp than from the center. Xenon lamp: If a xenon lamp is used, the difference in brightness is smaller than with a conventional fluorescent lamp, but this problem still exists. LED array: This is a strip of photodiodes. As all the diodes are equally bright, a shading plate is not needed.
1 April 2001
Page 199
Digital Processes
Image Processing
CCD
A CCD converts the light reflected from the original into an analog signal. The CCD (Charge Coupled Device) consists of a row of photosensitive elements. The circuit of each element in the CCD is shown at the right. Light hitting the photodiode charges up a capacitor. The brighter the light, the more charge goes into the capacitor. There is more about CCDs in the Standard Components chapter. The CCD has between 2,500 and 5,000 of these elements, depending on the maximum scanning width and number of pixels per unit length (i.e., the resolution across the page). A typical CCD in a high-end digital copier has 5,000 elements, at a resolution of 400 dpi (15.7 dots/mm).
c222d580.wmf
A CCD in a G3 fax machine may have a resolution of 8 or 16 pixels/mm, to match ITU-T standards. However, as many machines are now multi-functional, such machines often employ a dpi-based CCD and convert the signal to mm format when sending a Group 3 fax. The voltage from each element depends on the intensity of the light reflected from the original onto the element; the intensity of the light depends on the darkness of the area of the document it was reflected from. These charges are output from the CCD one after another, to make an analog video signal. Then the scanner moves to the next line of the original, and the CCD scans the next line. The CCD scans the original one line at a time, and outputs an analog signal for each line.
1 April 2001
Page 200
Digital Processes
Image Processing
Analog Signal Processing
Overview
Zeroing
Even
Signal Combining
Automatic Gain Control (AGC)
Z/C CCD Z/C
Odd
Black Level Analog Signal Input Z/C
Feedback Feedback
Feedback Peak Hold
A/D Converter 1 Ref 0 Ref Feedback
Digital Signal Output To Digital Processing Circuits
Black Level White Level
Auto Shading Circuits
ana-ads.wmf
1 April 2001
Page 201
Digital Processes
Image Processing
This section describes: · How the raw CCD output is prepared for conversion to digital data · How the corrected CCD output is converted to digital data The previous illustration shows the various steps and processes involved in preparing and converting the analog signal. The following table quickly summarizes each step. CCD output Auto shading Zeroing Signal combining Automatic gain control Black level Auto image density Peak hold A/D conversion How the raw data is output from the CCD. A key part of analog signal processing. It affects most of the other steps and processes. Black level correction prior to signal combination. Merging of the odd and even picture elements. Signal amplification and white level correction. Black level correction after automatic gain control. Removes background from the scanned image Holds the peak white value for A/D conversion. Conversion of the analog signal to a digital signal.
1 April 2001
Page 202
Digital Processes
Image Processing
CCD Output
This diagram shows the CCD and its data output lines as a simplified block diagram. There are two outputs from the CCD. One is for oddnumbered pixels, and the other is for even-numbered pixels. A clock switches the output for each pixel onto the odd or even output line alternately. Having two outputs speeds up the image processing. CCDs in older models (mainly fax machines) only had one output line. The two outputs are amplified before entering the analog signal processing circuits. Details about the amplification of the raw CCD output signal are given in section 8 (Components).
Photoelectric conversion CCD Reflected light
Even
Amplifier
ODD
Switching clock
Signal amplification
ccdblock.wmf
1 April 2001
Page 203
Digital Processes
Variations in the White Level
Image Processing
Auto Shading
Variations in the Black Level
a229d645.wmf
Auto shading corrects errors caused by variations in the signal level for each pixel. Both the black level and the white level are corrected.
1 April 2001
Page 204
Digital Processes
Image Processing
1) White Level Correction The video signal information for each pixel obtained during image scanning is corrected by the image processing circuits. The data has to be corrected for variations in white level across the page. These variations are caused by the following factors. · Loss of brightness at the ends of the exposure lamp with age or temperature (noticeable with fluorescent lamps and xenon lamps), or any bright and dull spots on the lamp · Less brightness at the edges of the lens · Variations in response among the CCD elements · Distortions in the light path, such as differences in reflectivity across the scanner mirrors. To correct for this, the machine scans a white plate before scanning each original. (This white plate is normally under the scanner cover or under the left scale of the exposure glass.) The white plate is uniform in color and in reflection. The output from each element of the CCD is converted to digital and passed to a memory in the auto shading circuit. The waveform of the white platen cover from the CCD is not uniform, because of the factors mentioned above.
c222d584.wmf
1 April 2001
Page 205
Digital Processes
Image Processing
In some models, there is a protection circuit which limits the white peak voltage. This is to prevent dark printouts resulting from an abnormally high reference voltage caused by strong light intruding into the scanner. In models that have a built in ADF, continuous scanning of large originals can cause the scanner to heat up, which affects the CCD's response. Also, continuous exposure to light affects the CCD. Therefore, the white plate is scanned every 30 s to recalibrate the white level (it is done between originals; scanning is not interrupted). After auto shading, the machine scans the page. The machine then uses the white waveform stored in the auto shading memory to correct the data. This is known as Automatic Gain Control (AGC). It is described later.
1 April 2001
Page 206
Digital Processes
Image Processing
2) Black Level Correction Method 1: Dummy Pixels This zeroes the black level for each scanned line of data while scanning the original. To get the current black level, the CPU reads the dummy data elements at one end of the CCD signal (some pixels at the end are blacked off), and takes an average of the voltages read from these elements. Then, the CPU deletes the black level value from each image pixel.
Output (V) Video Signal Before Correction Output (V) Video Signal After Correction
0 1 line
0 1 line
blk-lvl.wmf
This corrects the video signal for changes in response to the dummy black pixels as time passes. The black level is stored in the auto shading circuits (as a charge inside a capacitor, for example). Method 2: Black Level Waveform In some older models, the black level is done for every original, by shutting off the exposure lamp and reading a black level waveform across the page. This is stored in memory in the auto shading circuits in a similar way to that described earlier for the white level. Method 3: Fixed Reference Voltage Some models correct the black level using a standard reference voltage for the black reference (about 1.5 Volts)
1 April 2001
Page 207
Digital Processes
Zeroing
Even
Image Processing
Automatic Gain Control (AGC)
Signal Combining
Z/C CCD Z/C
Odd
Black Level Z/C Straight Through Analog Signal Input
Straight Through
Straight Through
Fixed Voltage Example: 2.5 V
A/D Converter 1 Ref 0 Ref
Digital Signal Output
Black Level White Level
Every line
From white plate, before each page Auto Shading Circuits
shadcct .wmf
When the machine scans the white plate before scanning the original, the odd and even pixel signals are combined. The resulting signal is converted to digital in the A/D converter, and stored in the memory in the auto shading circuits. The auto shading circuits are normally inside the digital processing circuits, and signals from this feed back into the analog circuits when needed. The black level goes to the auto shading circuit every line during scanning.
1 April 2001
Page 208
Digital Processes
Image Processing
Peak white can be detected every scan line too - this is Auto Image Density mode (also known as ADS mode). This is described later in this section. In the above diagram, the high level reference is arbitrarily fixed at 2.5 V and the low level reference at ground. In some cases, analog to digital (A/D) conversion is done using the peak value of the signal for the high reference, and half of the peak value for the low reference. Example: Model C211 The potential difference between the output of each pixel and the 53% level of the peak hold is converted by an A/D converter into 4bit data.
VPH Memory 100% 53% VT2100/2130/2150: 1.7V VT2300/2500: 1.4V 4 bits
5,000 pixels
shading1.wmf
1 April 2001
Page 209
Digital Processes
Image Processing
Zeroing
A zero clamp (Z/C) on each output adjusts the black level reference. The black level for the even pixels is adjusted to match the black level from the odd pixels. Feedback of the black level from the auto shading circuit is used.
Zeroing
Even
Signal Combining
Automatic Gain Control (AGC)
Z/C CCD Z/C
Odd
Black Level Z/C
Feedback Feedback
Feedback
Black Level White Level
Signal Combining
A multiplexer merges the analog signals for odd and even pixels from the CCD. In very high speed digital machines, the signals are not combined until the digital processing circuits. These machines have separate analog processing circuits for odd and even pixels.
Auto Shading Circuits
ana-ads.wmf
1
3
5
4995 4997
4999
1
2
3
4
4998 4999
5000
2
4
4996 4998
5000
vid-comb.wmf
1 April 2001
Page 210
Digital Processes
Image Processing
Automatic Gain Control (AGC)
The analog signal is amplified by operational amplifiers in the AGC circuit. When the original is scanned, the white level waveform is read back in from the auto shading memory. The AGC circuit uses the white level signal to correct the video data signal. In effect, each element of the scan line is amplified by an amount that depends on the voltage of the same element in the white level signal. An example is shown on the next page
CCD Z/C
Odd
Zeroing
Even
Signal Combining
Automatic Gain Control (AGC)
Z/C
Black Level Z/C
Feedback Feedback
Feedback
Black Level White Level
Auto Shading Circuits
ana-ads.wmf
1 April 2001
Page 211
Digital Processes
Image Processing
Element 500
For shading correction, the peak of the scan from the white plate is set to 1. Let us take an example, in which the level of the 500th element of the white waveform is 0.8 (i.e., not perfectly white). Then, at a point during scanning, say that element 500 in the video signal has a value of 0.6; it would be higher if there were no scanner irregularities. So, element 500 in the video signal is corrected as follows: 0.6/0.8 = 0.75. Each element in each video signal scan line is corrected in this way.
1 0.8 500 0 White W aveform Scan Line 0.6
500
0 Video Image Scan Line shadcorr.wmf
Also, if the platen cover is dirty, the values will be lower due to reduced reflection from the platen cover. This means that the image data will be overcorrected, causing pale bands in the image.
1 April 2001
Page 212
Digital Processes
Image Processing
Black Level
Before the data enters the A/D (analog-to-digital) converter, a zero clamp circuit again fixes the absolute value of the black level using feedback from the auto shading circuit.
Signal Combining
Automatic Gain Control (AGC)
Black Level Analog Signal Input Z/C
Feedback
Feedback Peak Hold
A/D Converter 1 Ref 0 Ref Feedback
Black Level White Level
Auto Shading Circuits
ana-ads.wmf
1 April 2001
Page 213
Digital Processes
Image Processing
Auto Image Density In some machines, this feature is called Original Background Correction. Auto Image Density (ADS) mode corrects for variation in background density down the page, to prevent the background of an original from appearing on copies.
ADS mode detects the background level for the original, also known as the peak white level, and removes this from the image, to make a white background. The machine must ensure that it detects white level from areas of the original that are free from image data. There are two methods, which are explained on the next page. When an original with a grey background is scanned, the density of the grey area becomes the peak white level density for that original. Therefore, the grey background will not appear on copies. Also, in machines where peak level data is taken for each scan line, ADS corrects for any changes in background density down the page. Unlike with analog copiers, the user can select a manual image density when in auto image density mode, and the machine will use both the manual and auto settings when processing the original. This is useful when making copies of an original that has light image density with background; AD removes the background, and if the user selected a dark manual image density setting, the image will be brought out more clearly in the copy.
1 April 2001
Page 214
Digital Processes
Image Processing
0.5mm
Method 1: Scanned from a narrow strip near the rear scale (Example: Model A229) The copier scans the auto image density detection area [A]. This corresponds to a narrow strip at one end of the main scan line, as shown in the diagram. As the scanner scans down the page, the machine detects the peak white level for each scan line, within this narrow strip only. Method 2: Scanned from a narrow strip at the center of the leading edge (Example: C211 series)
[A]
15mm 75mm
Sub scan direction
a229d581.wmf In this machine, the original is placed at the center of the original feed path, and not at one side like in the A229. Therefore, the peak level is read from the central 64 mm at the leading edge of the original.
One problem with this method is that, since scanning starts before the light intensity from the fluorescent lamp stabilizes, the light intensity tends to increase for a little while. The voltage from the CCD increases until the light intensity stabilizes. As a result, lighter image densities may not appear on prints after the light stabilizes. To prevent this, the peak voltage is changed when a higher (whiter) image signal is detected. If the peak voltage changes regardless of the output value, like in the A229, there is a chance of mistaking grey areas in the center of the image for peak white.
1 April 2001
Page 215
Digital Processes
Image Processing
Automatic Gain Control (AGC)
The peak hold circuit holds the peak white level. From this peak white level, the machine determines the white reference value for A/D conversion. The white level from auto shading is fed back to the ADS circuit to correct for fluctuations in the white level across the page.
Signal Combining
Black Level Analog Signal Input Z/C
Feedback
Feedback Peak Hold
A/D Converter 1 Ref 0 Ref Feedback
Black Level White Level
Auto Shading Circuits
ana-ads.wmf
1 April 2001
Page 216
Digital Processes
Image Processing
A/D Conversion
The A/D converter converts the analog signal to digital. In a typical machine, the resulting digital signal has eight bits. This means that each pixel can have one of 256 values. However, before this can be done, the A/D converter must be supplied with reference voltages that determine the black and white limits. To do this, the A/D converter is supplied with a black reference voltage (0 Ref). For example, the input could be held to ground. This fixes the lowest of the 256 levels any pixel with the same voltage as the black level will become black. Also, the highest of the 256 values is fixed with a white reference voltage (1 Ref). When the analog signal is digitised, 0 Ref and 1 Ref will serve as references for black and white, and the 256 levels of the grey scale will be distributed between these two levels. If ADS is not being used, the white reference (1 ref in the diagram) is held to a fixed voltage.
Analog Signal Input
Fixed Voltage Example: 2.5 V
A/D Converter 1 Ref 0 Ref
Digital Signal Output To Digital Processing Circuits
ananoads.wmf
1 April 2001
Page 217
Digital Processes
Image Processing
If ADS is being used, the white reference voltage depends on the output of the peak hold circuit.
Signal Combining
Automatic Gain Control (AGC)
Black Level Analog Signal Input Z/C
Feedback
Feedback Peak Hold
A/D Converter 1 Ref 0 Ref Feedback
Black Level White Level
Auto Shading Circuits
ana-ads.wmf
The A/D converter divides the range between the black and white reference voltage into 256 levels and digitizes the analog signal based on these levels. These 256 levels are known as grayscales. The low reference voltage terminal stays constant. Only the high reference terminal voltage varies.
1 April 2001
Page 218
Digital Processes
Image Processing
Analog 2.9 V 256 Digital ........................... 00000000 White
Example: Model A099 In this example, the signal has been inverted so that digital 0 is white and 1 (0 Volts) is black. The white level varies between 1.7 and 2.9 V, depending on the feedback from the peak hold circuit for ADS. (If ADS was not being used, the white level would remain fixed.) The A/D converter divides up the range from black to the current white level into 256 levels. The grey scale is based on the peak white level. The right side of the diagram shows how the range is divided up if the white level is 1.7 V. If the white level was 2.9 V, the spacing would be wider.
0Ref Range
1.7 V 256 255
............................ 00000000 ............................ 00000001 256 levels calculated as follows: 256 0Ref (D is the Digital data) D= Vin x
4 3 2 1 0V
............................ 11111100 ............................ 11111101 ............................ 11111110 ............................ 11111111
Black adcon.wmf
If the voltage for a pixel is between level 2 and level 3, this is converted into a digital value of 11111101. Pure black (above level 255) becomes 00000000. Pure white (below level 1) becomes 11111111.
1 April 2001 Page 219
Digital Processes
Image Processing
Digital Signal Processing
Overview This section explains how the raw digital data from the A/D converter is processed to produce a faithful image of the original. Digital fax machines, scanners, printers, and copiers use a wide range of digital image processing tools. The processes used are different in every machine, and so is the order in which they are done. Because of this, a comprehensive description is impossible. However, representative examples will be given. Many of the processes are proprietary, and in these cases, details cannot be given.
Digital processes can be broadly classified into the following types. ! Preliminary Image Enhancement: These processes prepare the data for processing by correcting the data for scanner characteristics, and removing unwanted data such as dots in the background. · · · · Scanner Gamma Correction Background Erase Independent Dot Erase Text/Image Separation
1 April 2001
Page 220
Digital Processes
Image Processing
! Filtering: These processes enhance the data to suit the original mode (text or photo) selected by the user. · · MTF (Modulation Transfer Function) Photo mode Smoothing
! Magnification and Reduction: This enlarges and reduces the data, depending on the reproduction ratio selected by the user, or the paper size in the receiving fax terminal. ! Gradation Processing: The gradation processing methods used generally depend on the original type setting (text, photo, etc) selected by the user. · · · · Grayscale Processing Binary Picture Processing Dithering Error Diffusion
! Editing and Merging Using a memory work area, digital data can be manipulated to produce various effects, such as combining several images onto one copy. Also, multiple originals can be scanned into memory and several copies can be printed, already sorted, onto a single output tray. This is sometimes called electronic sorting. This feature allows low-volume sorted output without needing all the extra hardware.
1 April 2001
Page 221
Digital Processes
Image Processing
Another benefit of digital processing with memory storage is faster duplex copying throughput, using a feature known as 'interleaving'. This feature uses a duplex tray with a one-page capacity, stores multiple originals in memory, and outputs the data in the order that is suitable for the fastest printing. This order is not necessarily the order in which the pages were scanned. This is covered more fully in the Paper Handling section (Interleave Duplexing). The main benefits for most users are that a job with multiple originals can be scanned just once and stored in memory, then printed many times from memory without having to scan again. Also, printer jams can be recovered without having to scan the original again. · · · · · · · · · Merging Make-up Mode Image Rotation Combining Images Erasure of Irregular Dots Line Width Correction Edge Detection Sub-scan Resolution Conversion Inch-mm Conversion
! Final Image Enhancement
1 April 2001
Page 222
Digital Processes
Image Processing
Scanner Gamma Correction
Scanner gamma correction corrects the data to account for the characteristics of the scanner (e.g., CCD response, scanner optics). This ensures that the various shades in the grey scale from black to white on the copy match those on the original . The relationship between original image density and analog circuit output should be linear as shown in the upper diagram. However, in reality, it is more like that shown in the lower diagram. Gamma correction corrects the data for this deviation, as shown by the arrows in the lower diagram.
Scangam1.wmf
1 April 2001
Page 223
Digital Processes
Image Processing
In some machines, the gamma curve can be changed with a service mode. Also, some machines automatically adjust the gamma curve depending on the image density setting selected by the user. Example 1: Model C222
Dark image setting
Normal image setting
C222D588.wmf
If the user selects `dark' mode, the `dark image' gamma curve is used and the output is darker.
1 April 2001
Page 224
Digital Processes
Image Processing
Example 2: Model C210 In this machine, there are four different image density settings, as shown (Darker 2, Darker 1, Normal, Lighter), There is an additional adjustment for tone. Using these, the user can emphasize better reproduction of pale or dark tones. For example, if the user selects `dark tone' mode (solid lines), the gamma curves change so that the output changes rapidly for small changes in input at the dark end of the scale. (The dotted lines show the curves for normal tone.) This causes shades of grey at the dark end of the scale to be reproduced. There is also a printer gamma correction, to adjust the data for printer characteristics. This is discussed in the Laser Exposure section.
Output
Dark Tone Mode
Darker 2
Darker 1 Normal
Lighter
Normal Tone
Input
na2gamma.wmf
1 April 2001
Page 225
Digital Processes
Image Processing
Background Erase
Usually, dirty background is erased using Auto Image Density (ADS). However, sometimes, dirty background areas will still appear. These can be erased by Background Erase. If any low image density data which is lower than a threshold level remains after auto shading, this data will be treated as '0', which is equal to 'White'. By adjusting the threshold to a larger value, darker backgrounds can be eliminated. Example: Model A229 If there is a sudden cutoff at the threshold, sudden changes in the data around the threshold level area can cause errors during the MTF process. So, in the example shown, the image density does not cut off at the threshold [A], but gets paler more rapidly than usual, until at a certain point [B] it becomes white.
Output
255
0 [B] [A]
255
Input
A229D591.WMF
1 April 2001
Page 226
Digital Processes
Image Processing
Independent Dot Erase
This feature removes isolated black pixels from the image. It is normally not used in photo mode, to avoid deleting details from images. Example: Models A230/A231/A232 The software compares each pixel (C in the diagram above left) with the pixels around the edges of the surrounding 3 x 5 area. If the sum of the pixels at the edges is smaller than the threshold value, the object pixel is changed to 0 (white) or reduced in density to an average of the pixels around the edge, depending on an SP mode setting. The threshold can also be adjusted. In the example shown to the right, if the pixel is below the threshold value, it is either erased, or reduced to 3 (the average of the pixels around the edge, which is 37 divided by 12).
A1 A6 A8 A2 A3 C A4 A5 A7 0 0 0 0 0 30 90 0 0 7 0 0 0
Original image
A9 A10 A11 A12
3 x 5 area
Image data
A231D528.WMF
1 April 2001
Page 227
Digital Processes
Image Processing
Text/Image Separation
When the user selects Text/Photo mode, the machine processes text areas and image areas differently. Some machines have only a simple text/image separation as part of the error diffusion process (described later), whereas others have a more sophisticated algorithm (described in this section). Note that for some machines, "Letter mode" is used to refer to originals containing text. "Letter" refers to a type of image data, not Letter size paper or "correspondence". It means text and/or line art. Method 1: Edge and dot screen area detection Generally, text areas have strong contrast between the image and the background. In photo areas (dot screen areas), there is a less extreme range of contrast, and mid-range grey areas are common. By using these characteristics and the following separation methods, the original image is separated into text and photo areas. 1. Edge detection
Text/Photo Separation
Edge Determination Final Evaluation
Dot Screen Determination
a229d625.wmf
Edges of letters and parts of images are detected by checking for strong contrast, continuity of black pixels, and continuity of white pixels around the black pixels.
1 April 2001
Page 228
Digital Processes
Image Processing
2. Photo area (dot screen) detection Each pixel is tested to see if it is in a dot screen area by comparing with nearby pixels. Example: Model A229 The page is divided into 4 x 4 blocks of pixels. Each block is placed at the center of a 5 x 3 array of these blocks, and becomes either text or photo, depending on the other blocks in the 5 x 3 area .
Dot Screen Dot Dot Screen Screen Dot Dot Dot Screen Screen Screen Dot Screen
If the number of dot screen blocks in the 5 x 3 area exceeds a threshold, the central Determined to be Photo block is determined to be an image area. (The threshold is 2: if two or more of the blocks in the 5 x 3 area are dot screen, areas then all the pixels in the central block are determined to be in an image area.)
Determined to be Text
a229d640.wmf
1 April 2001
Page 229
Digital Processes
Image Processing
Final Evaluation The machine decides whether each pixel is in a text or image area by looking at the results of the edge and dot screen detection processes. Example: Model A229
Dot Screen No No Yes Yes Edge No Yes No Yes Final Evaluation Photo Text Photo Photo
GA3 Filter
MTF Correction Selector Switch Smoothing
Text and image areas can then be processed differently. Example: Model A133 The image data is treated by MTF and by smoothing simultaneously. However, the result of the final evaluation controls a selector switch. For a text area pixel, the output from the MTF selector is selected. For an image area pixel, the output from the smoothing circuit is selected.
Auto Text/Photo
Edge Detection Final Evaluation Dot Screen Detection
a133d549.wmf
1 April 2001
Page 230
Digital Processes
Image Processing
Method 2: Comparison of adjacent pixels Example: Model C226 In the Letter/Photo mode, the machine checks each pixel of the original to see if the pixel is in a line area or in a photo area. To recognize a line area in a photo original, the CPU does the following calculation on the 6-bit pixel data. x = | (c + f + i) - (a + d + g) | y = | (g + h + i) - (a + b + c) | If x or y is greater than 10, the machine recognizes that pixel e is in a letter area. If the calculated number is 10 or less, the pixel is in a photo area. In larger digital machines, this is a part of the error diffusion process, in addition to the main text/image separation process described earlier.
c222d595.wmf
1 April 2001
Page 231
Digital Processes
Image Processing
MTF (Modulation Transfer Function)
When the CCD converts the original image to electrical signals, the contrast is reduced. This is because neighboring black and white parts of the image influence each other as a result of lens characteristics. This symptom is typical when the width and spacing between black and white areas are narrow. MTF correction counters this symptom and emphasizes image detail. Because of this, MTF is necessary for reproduction of details such as thin lines, points, and complex characters. Without MTF, such details may be lost, or only partly reproduced. Small dots and thin lines may be split up over more than one pixel. If the dot or line is small enough, the pixel output may fall below the threshold required to register a black pixel, and it would not be printed. Because MTF sharpens the image, it is normally not used with photo mode. However, MTF can be useful in photo mode when putting more weight on improving the resolution when copying from continuous tone originals. Also, in text/photo or photo mode, MTF can be combined with error diffusion, which reduces differences in contrast. The MTF algorithm generates a new value for the density of the element, using an algorithm that uses the density values of neighboring pixels in the image.
1 April 2001
Page 232
Digital Processes
Image Processing
Example: Model C223 Consider a small black point on a original as shown in the illustration (a) and (b). The 6-bit image data (range 0 to 63) for this section of the original is shown in (c). If the threshold level is 32, all the pixels in this area will become singlebit white data and the image will not be reproduced (d). The MTF correction prevents this image loss by modifying the value of each pixel in the following manner
a) Section of original
b) Enlarged view of dot
0 0 0 0 0
0
0
0 0
12 4
30 12 0 0 0 0 0 0 0
d) Print without MTF correction (threshold level: 32)
c) Image data after A/D conversion
The value of the target pixel is multiplied by 3. Then, 3/8 of the values of the pixels to the left and right, 1/8 of the values of the pixels two steps to the left and right, and 1/2 of the values of the pixels above and below are subtracted from the new value of the target pixel. (If the result is less than zero, then the pixel value is set to zero.)
C223d667.wmf
-1/2 -1/8 -3/8 3 -3/8 -1/8 -1/2
c223d668.wmf
1 April 2001
Page 233
Digital Processes
Image Processing
0 0 0 0
After the MTF correction is applied, the image data of the example is as shown in (e) and (f). The small black point is reproduced on the print.
0 19.5 1.5 0 0 0 0
63 22.7 0
0 0
0 0
0 0
f) Printout after MTF correction
e) Image data after MTF correction
c223d624.wmf
The MTF algorithm can be strengthened by using higher values in the calculation. See the example on the right. In some machines, the MTF algorithm can be strengthened in either the main scan direction, sub scan direction, or both at once. For example, if the original has a lot of thin horizontal lines, MTF can be strengthened in the sub scan direction to preserve these lines, without applying an excessive MTF in the main scan direction.
-1/8 -3/8
-2 6 -2
-3/8 -1/8
C223d625.wmf
A stronger MTF filter sharpens the image and leads to better reproduction of low image density areas, but may lead to the occurrence of moiré in the image. Also, stains, scratches, and other blemishes in the light path will appear on prints more easily.
1 April 2001
Page 234
Digital Processes
Image Processing
Photo mode smoothing
There are some different processes that use the name 'smoothing'. This section describes the image enhancement process that is used in photo mode to make a softer image. The other types of smoothing act on the final data to remove jagged edges from the image. They will be described later. Smoothing acts in a directly opposite way to MTF. It smoothes the contrast between adjacent pixels, giving better reproduction for photos. Because of this, it will not normally be used in text mode.
1 April 2001
Page 235
Digital Processes
Image Processing
Example: Model A099 The smoothing algorithm is: the values of the 24 pixels surrounding the object pixel and the object pixel are multiplied by the values in a 5x5 filter matrix. Then the new values are added together. The result is then divided by 64 and rounded off to yield the new value of the pixel. If this procedure is applied to the example, the value of the pixel shown in the figure changes from 18 to 17. This algorithm is applied to all pixels. If the pixel is on the edge of the image area, the missing data is assumed to be "0".
5
14 14 14 14 18 14 14 14 18 19
5
14 14 18 18 19 14 18 19 20 21 18 18 19 20 21
Image
: Object Pixel
1 2 2 2 1
2 4 4 4 2
2 4 4 4 2
2 4 4 4 2
1 2 2 Filter 2 1 (1/64)
17
Result
c4smth-1.wmf
1 April 2001
Page 236
Digital Processes
Image Processing
The filter can be changed using a service program to suit the type of original. Example: Model A099 again High-contrast originals
1 1 2 1 1 1 2 2 2 1 1 1 1 1 1
2 4 4 4 2 2 4 4 4 2 1 1 1 1 1
2 4 8 4 2 2 4 4 4 2 1 1 1 1 1
2 4 4 4 2 2 4 4 4 2 1 1 1 1 1
1 1 2 1 1 1 2 2 2 1 1 1 1 1 1
Page 237
Normal originals
Low-contrast originals
c4smth-2.wmf
1 April 2001
Digital Processes
Image Processing
Magnification and Reduction
Overview If the user selects a magnification or reduction ratio at the operation panel before copying, the image data must be enlarged or reduced. Also, fax machines have to reduce the data if the paper in the machine at the other end is not wide enough to print the message. The machine determines whether reduction is necessary by comparing the received protocol signal with the document width sensor readings. Sub Scan Direction Method 1: Original transport speed Example: Model A229 Reduction and enlargement in the sub scan direction are done by changing the scanner or ADF motor speed.
1 April 2001
Page 238
Digital Processes
Image Processing
Method 2: Deleting scan lines The cpu does sub-scan reduction by cutting out the 3rd and 7th scan lines in every 7 scan lines (for A3 [11.7" x 16.5"] to A4 [8.3" x 11.7"]), or the 6th and 13th scan lines in every 13 scan lines (for A3 [11.7" x 16.5"] to B4 [10.1" x 14.3"] and B4 [10.1" x 14.3"] to A4 [8.3" x 11.7"]). This is only done by older fax machines. Recent models change the scanner motor speed. Example: A3 to A4 reduction Main Scan Direction Reduction and enlargement in the main scan direction are handled by digital image processing circuits. Method 1: Calculation of Imaginary Pixels
faxsubrd wmf
1 April 2001
Page 239
Digital Processes
Image Processing
Example: Model A229
Scanning and laser writing are done at a fixed pitch (the CCD elements cannot be squeezed or expanded). So, to reduce or enlarge an image, imaginary points are calculated that would correspond to a physical enlargement or reduction of the image. The correct image density is then calculated for each of the imaginary points based on the image data of the nearest four true points. The calculated image data then becomes the new (reduced or enlarged) image data.
1 April 2001 Page 240
Digital Processes
Image Processing
80 % Reduction For example, data for 10 pixels in a main scan line are scanned by the CCD. Those data are compressed into data for 8 pixels by the magnification processor. As a result, the image is reduced to 80 %. 140 % Enlargement Data for 10 pixels of a main scan line are expanded into data for 14 pixels. As a result the image is enlarged with a 140 % magnification ratio. The calculation method is described below in more detail.
1 April 2001
Page 241
Digital Processes
Image Processing
To reduce or enlarge an image, imaginary points are calculated that would correspond to a physical enlargement or reduction of the image. The image density is then calculated for each of the imaginary points based on the image data of the nearest four true points. The calculated image data then becomes the new (reduced or enlarged) image data. Here is an example of how the calculation is done.
2 Scanned Data Point 2' Calculated Data Point
3
4
5
3'
4'
r
1-r
1+r
2-r
Mainscan.wmf
In the example on the right, the density at point 3' (3') is calculated from the densities at points 2, 3, 4, and 5 (2, 3, 4, and 5) as follows: (3') = 2 x h(1+r) + 3 x h(r) + 4 x h(1-r) + 5 x h(2-r) h(1+r) + h(r) + h(1-r) + h(2-r) The values of the weighting factors h(1+r), h(r), h(1-r), and h(2-r) depend on the value of r, as shown in the table on the right. The set for the nearest value of r is used. r 0 0.25 0.5 0.75 h(1+r) 0 - 0.25 - 0.25 - 0.25 h(r) h(1-r) 1 0 1 0.375 0.75 0.75 0.375 1
h(2-r) 0 - 0.125 - 0.25 - 0.25
1 April 2001
Page 242
Digital Processes
Image Processing
Method 2: Adding and Deleting Pixels Another way to expand or shorten the main scan line is to add or delete pixels at regular intervals. However, this method is not so flexible as method 1, because it does not allow the user to increase or decrease the magnification in 1% steps (the `zoom' feature). Example: Model C226 Reduction Mode 100% 93% 82% (A4 version) 75% (LT version) 71% (A4 version) 64% (LT version) Discarded Pixels 0 Pixels 1/14 Pixels 3/11 Pixels 1/4 Pixels 2/7 Pixels 5/14 Pixels Remaining Pixels All Pixels 13/14 Pixels (0.929) 9/11 Pixels (0.818) 3/4 Pixels (0.75) 5/7 Pixels (0.714) 9/14 Pixels (0.642)
71% reduction: 5 out of 7 pixels are used, 2 pixels are discarded (see the diagram). 82% reduction: 9 out of 11 pixels are used, 2 pixels are discarded. 93% reduction mode: 13 out of 14 pixels are used, 1 pixel is discarded.
71red.wmf
In some machines, there is one exception to this rule. If the
1 April 2001 Page 243
Digital Processes
Image Processing
pixel scheduled for deletion is darker than the pixel immediately to the right, the latter pixel is deleted instead. Enlarge Mode 115% (LT/A4 Version) 122% (A4 Version) 127% (LT Version) 141% (LT/A4 Version) Added Pixels 2 Pixels 3 Pixels 3 Pixels 9 Pixels Pixel Ratio 15/13 Pixels (1.154) 17/14 Pixels (1.214) 14/11 Pixels (1.273) 31/22 Pixels (1.409)
115% enlargement mode: Every 7th pixel and 13th pixel are doubled to produce 15 pixels from every 13 pixels in the original (see the drawing). 122% enlargement mode: Every 5th, 10th, and 14th pixels are doubled to produce 17 pixels from every 14 pixels in the original. 127% enlargement mode: Every 4th, 8th and 11th pixels are doubled to produce 14 pixels from every 11 pixels in the original. 141% enlargement mode: Every 3rd, 5th, 8th, 10th, 13th, 15th, 18th, 20th, and 22nd pixels are doubled to produce 31 pixels from every 22 pixels in the original.
1 April 2001 Page 244
115enl.wmf
Digital Processes
Image Processing
Some digital processes can cause moiré when used in conjunction with reduction or enlargement at certain reproduction ratios. Because of this, the order of some processes depends on the reproduction ratio. Example: Model A229 64% reduction or less: Main Scan Reduction then Filtering (MTF or Smoothing) 65% reduction or higher: Filtering (MTF or Smoothing) then Main Scan Magnification/Reduction
1 April 2001
Page 245
Digital Processes
Image Processing
Method 3: Laser Diode Pixel Width Example: Model H523 The CPU controls the magnification ratio by changing the interval between pulses in the laser clock signals. So, for example, the clock signal pulse interval for 200% enlargement is twice as long as the interval for normal (100%) image reproduction. This makes each image pixel for 200% enlargement become twice as long as each pixel for normal image reproduction.
1 April 2001
Page 246
Digital Processes
Image Processing
When using the ADF, the magnification circuit has to create a mirror image. This is because the main scan starts is at the other end of the scan line in ADF mode (as compared with platen mode). In platen mode, the original is placed face down on the exposure glass, and the corner at [A] is at the start of the main scan. The scanner moves down the page. In ADF mode, the ADF feeds the leading edge of the original to the DF exposure glass, and the opposite top corner of the original is at the main scan start position. To create the mirror image, the CPU stores the main scan line data in a LIFO (Last In First Out) memory from the last pixel. When loading the main scan line data from the LIFO memory, the CPU loads the first pixel of the main scan line.
A193D504.wmf
1 April 2001
Page 247
Digital Processes
Image Processing
Moiré
When one pattern is imposed over another sometimes they interfere with each other and form a third pattern called a moiré pattern. In our products, MTF processing is a major cause of moiré patterns.
CCD Original Image Data for one CCD element CCD Output The illustration shows one of the moiré Waveform Data mechanisms. In this case, the pixel density 1.7 V of the CCD is the same as the density of the regular lines on the original. However, C223d635.wmf the regular lines are slightly out of step with the CCD pixels. As a result, each CCD pixel has a different value (as shown in the figure). Since the length of a CCD pixel is very short, the waveform from the CCD output looks like the cross lines in the figure. The moiré pattern appears when prints are made from this signal. The moiré pattern typically appears when the CCD pixel density is a multiple of the density of the regular lines on the original.
1 April 2001
Page 248
Digital Processes
Image Processing
Grayscale Processing
Grayscale processing uses many shades of grey to reproduce continuous tone originals, such as those containing photographs. A black and white photograph contains an unlimited number of shades of grey, but digital copiers and printers can normally only output a few shades, normally 64 or 256. If grayscale processing is used, the digital image processing circuit outputs, to the memory or laser diode driver, the result of all the previous enhancement and filtering processes, without any error diffusion or dithering. The result is a multi-bit per pixel stream of digital data. For example, if there are 256 shades of grey, there are eight bits per pixel. Note that grayscale processing needs a lot of memory. At eight bits per pixel (256 shades of grey), an A4 or LT page needs about 14 megabytes, without compression.
1 April 2001
Page 249
Digital Processes
Image Processing
Binary Picture Processing
In binary picture processing, the output data is one-bit only. There are no shades of grey. the output is black or white only. The multi-bit per pixel data stream has to be reduced to single-bit data. To do this, a threshold level is used. If a pixel has a value that is brighter than the threshold, it becomes a white pixel. If it is darker than the threshold, it becomes a black pixel. The threshold can usually be adjusted, and it often varies depending on modes selected at the operation panel. The example on the right shows how the threshold level affects the output. If binary picture processing is used with dithering or error diffusion, then the threshold level for each pixel will be different, as described in later sections.
Output 60 36 24 20 Output 60
20 16 8
Density Original Original
Density
Print
Print
Threshold Level: 24
Threshold Level: 16 thresh.wmf
1 April 2001
Page 250
Digital Processes
Image Processing
Dithering
This is used to reproduce originals with continuous tones, such as photographs on machines that cannot output true grayscales. Dithering produces different shades of gray by making different patterns of black and white dots. There are no gray dots at all. Dithering is sometimes called halftoning, and the various shades of gray are called halftones.
1 April 2001
Page 251
Digital Processes
Image Processing
Example: Model C211
Main Scan
5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 4 14 9 11 7 0 5 4 1 2 4 8 3 6 14 9 7 0 4 1 2 8 3 6
Sub Scan
11 5
5 5 5 5 5 5
12 13 10 10 12 14 14 13 8 7 4 4 8 14 14 7 1 3 3 9 0 0 1 9 6 11 5 2 2 6 11 5 12 10 14 13 10 12 14 13
Image Data 4-bit data All pixels at 5
Dither Matrix
Result
dith1.wmf
The diagram shows how a dither matrix is used. In this machine, a 4 x 4 dither matrix is used, repeated many times so that it becomes the same size as the data for the scanned original. The dither matrix contains threshold levels. Each pixel of the scanned image is compared with the threshold level at the same location in the dither matrix. Then, each pixel changes to either black or white depending on whether the image data is greater or less than the threshold level. This procedure is repeated for the whole of the original. In the example, the original is a single tone of grey, and the repeated pattern output from the dither matrix appears grey to the human eye.
1 April 2001 Page 252
Digital Processes
Image Processing
The thresholds in the dither matrix are designed so that half-tones can be reproduced on prints using only black and white pixels, by changing the ratio of black pixels to white pixels. The matrixes can be adjusted in many machines to increase or decrease the detail on the copy. Also, the greater the number of lines in the matrix, the better the image quality in photo mode. Example: Coarse and Fine Screen Mode in Model C223 In this model a 12 x 12 dither matrix is used to convert 8-bit image data into single-bit data. The dither matrix for fine screen mode is different from the one for coarse screen mode. The diagram shows what happens to an original with a constant grey tone of grade 55 (out of the possible 256).
1/400 inch
dith2.wmf
Fine Screen
Coarse Screen
1 April 2001
Page 253
Digital Processes
Image Processing
Error Diffusion
The error diffusion process reduces the difference in contrast between light and dark areas of a halftone image. Each pixel is corrected using the difference between it and the surrounding pixels. The corrected pixels are then compared with an error diffusion matrix. Compared with dithering, error diffusion gives a better resolution, and is more suitable for "continuous toned" originals. On the other hand, dithering is more suitable for "screen printed" originals. Error diffusion is often used in text/photo mode. Dithering reproduces text areas poorly, and with just a simple thresholding or grayscale process, photo areas do not come out well. Error diffusion is a good compromise because it reduces the contrast between light and dark areas of halftone images, while having no effect on letter areas. Example: Model C226 Before a 6-bit image signal is converted into a single-bit signal based on the threshold level, there is a difference between the image signal value and the complete black value (63 for a 6-bit signal) or white value (0). With the Error Diffusion process, the difference is distributed among the surrounding pixels. (The MTF process simply erases these differences.) When considering Error Diffusion in one dimension only (across the page), the 6-bit data shown in the example below produces white and black data output as shown below. In practice, this onedimensional Error Diffusion is done in all directions on each pixel (across the page, down the page, etc.).
1 April 2001
Page 254
Digital Processes
Image Processing
In each dimension, the difference between the pixel value and the nearest extreme (0 or 63) is transferred to the next pixel. The 1st pixel in the row becomes either black or white, whichever is closest. Then, for the 1st pixel above, the difference between 7 and 0 is added to the 2nd pixel. The value of the 2nd pixel, which is now 18, is then added to the 3rd pixel. The 4th pixel becomes 52, which is closer to 63 than 0. In such cases, the difference is subtracted (not added) to get the next pixel value. In this example, the difference is 63-52=11, and the next pixel value (30-11) becomes 19. These values will then be treated by an error diffusion matrix.
na2erdif.wmf
c222d590.wmf
1 April 2001
Page 255
Digital Processes
Image Processing
In Text/Photo Mode, the error diffusion matrix that is used may depend on the image area type (text or photo). Therefore, before error diffusion, a simple text/photo separation process is performed. This was described in Text/Image Separation - Method 2: Comparison of adjacent pixels. If error diffusion is used with binary picture processing, the output image signal level has just 2 levels (white and black). If it is used with grayscale processing, the output image signal level has a number of levels (from white to black). For example, in a machine with 256 grayscale output, the output from error diffusion may use a small selection of these values, which are selected to give a good print quality. Example: Model A229 (256 grey scales) Photo mode 17 levels per pixel Text areas in text/photo mode 9 levels per pixel Photo areas in text /photo mode 17 levels per pixel
1 April 2001
Page 256
Digital Processes
Image Processing
Merging
Digital processing allows the user to combine other forms of data with the original before printing. Common examples include printing the date and time, printing a message (such as 'Confidential'), or printing a background pattern.
Make-up Mode
In make-up mode, the user scans command sheets before the original. Each command sheet specifies an area of the original. Before making the copy, the user then specifies which effects to use for the designated areas of the original. Typical effects include original type (text, photo, etc), use of various colored inks, reversed image, and background patterns. This is a common feature in Priport machines. Color copiers achieve something similar using an editor touch screen on the operation panel.
Image Rotation
If the machine has paper of the same size as the original but different orientation, the image will be rotated by 90 degrees in memory before printing. The machine must have enough working memory to do this. The amount of memory required for a certain paper size depends on the image resolution and the number of bits per pixel. In the A229, 12 MB of DRAM is enough to hold two A4 images. This allows users to scan one original into the RAM while still copying from another. This only works for originals up to A4.
1 April 2001
Page 257
Digital Processes
Image Processing
Combining Images
Using the memory, digital machines can print reduced images up to eight pages on one sheet of copy paper, or 16 pages using duplex mode. If the locations of the printed images are arranged suitably, the user can make a small booklet out of up to 16 originals, using duplex mode, then folding and cutting the copy. Example: Model A133
comb1 wmf
comb2.wmf
1 April 2001
Page 258
Digital Processes
Image Processing
Erasure of Irregular Dots
After the binary picture processing stage, some fax machines use pattern recognition to remove irregular dots. Example: Model H515 If an element after being converted to white or black by binary picture processing is irregular against the surrounding pixels, it is output in the opposite color. The central pixel is compared with the surrounding eight pixels to determine whether this process is necessary. There are ten cases, as shown above, in which conversion is done. This results in a cleaner image.
H515D647.wmf
1 April 2001
Page 259
Digital Processes
Image Processing
Line Width Correction
This is normally only used when the user selects Copied Original mode (making a copy of a photocopy). This mode is known as `Generation Copy' mode in some machines. Some copiers cause lines to bulge in the main scan direction as a result of the development system. So, pixels on edges between black and white areas are compared with adjacent pixels, and if the pixel is on a line, the line thickness will be reduced. For example, if the line on the original is one pixel in width, the pixel on the copy may be slightly larger than one pixel width (as shown in the bottom diagram) due to the shape of the dot made by the laser beam and the amount of toner attracted to the pixel. If this copy is used as an original, image processing may then generate additional black pixels at the edges of the line, so the resulting printout will have slightly thicker vertical lines in places. Line width correction attempts to correct for this effect.
Original Copy
Original
Copy
Lwc.wmf
1 April 2001
Page 260
Digital Processes
Image Processing
Edge Detection
In some fax machines, this process preserves the sharpness of image outlines. Each element is tested to determine whether it is on a boundary of two areas of sharp contrast (such as the edge of a character on a white background). If the element is on a boundary, it goes straight to the cpu as a black (1) element. (Halftone processing on this element could lead to a fuzzy outline.) Edge detection uses a threshold, which can be adjusted by RAM address in some models if edges of characters appear fuzzy.
Sub-scan Resolution Conve