 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of Invention
The present invention relates to a digital image processing apparatus, and
more particularly to compression of the image data.
2. Description of Related Art
In recent years, digital photocopiers which generate a hard copy of a
manuscript by reading the manuscript image using an image input apparatus,
for example a scanner, digitally processing the input image data, and
outputting the digitally processed data to an image output apparatus, for
example a printer, have come into widespread use.
In the digital photocopier, it is essential to store a plurality of the
image data in the photocopier, have an electronic sorter function which
sorts the data (for example, manuscript, files, and edits pages), and to
have an electronic RDH function. This is accomplished by equipping the
copier with an interval data storage apparatus, for example, random access
memory or a hard disk, storing the image data therein and outputting the
data as necessary. In order to store a large amount of image data, it is
necessary to increase the storage capacity of the storage apparatus, but
this results in an increase in the size and cost of the apparatus itself.
In order to avoid this, a method of storage that compresses the image data
has been proposed. By compressing the image data, it is possible to store
a large amount of the image data using smaller-capacity storage apparatus.
Furthermore, the image output apparatus maybe a laser printer. In general,
a page descriptive language is used for controlling the method of image
output to the printer. A host computer, to which the printer is connected,
does not transfer to the printer the output content itself as a bitmap
image (raster image) but rather the content of the page descriptive
language describing the character and image information of that output
content. The printer receives the page descriptive language, internally
interprets the language content, render the image data of the page as a
bitmap image (raster image), and outputs the image by transferring the
image to the paper.
Therefore it is essential to have sufficient printer memory to maintain the
bitmap image which renders the image data and a function to interpret the
content of the page descriptive language. For example, memory would
require 32 megabytes if the output resolution is 400 dpi and the output
gradation degree is 256 steps, in the case of a monochrome printer which
outputs A3 size paper.
In the case of a color printer, the output of YMCK 4 colors is needed, so
the memory capacity becomes four times larger, to 128 megabytes. Obtaining
such large memory capacity necessarily increases the size, cost, etc., of
the printer itself. To avoid this increase in memory capacity, just like
the case of the digital photocopier, the image data can be stored in the
compressed format. In doing so, large amounts of image data can be stored
using smaller memory capacity.
In the case of compressing the image data and reducing the data capacity,
reducing the gradation degree of the image repeatedly and storing the
image in a binary state can be considered, however, the quality of the
image output which can be eventually obtained deteriorates when the
gradation degree is reduced. Thus, in order to store images of high
quality, it is better to store the image in a multi-value state rather
than in a binary state.
In order to compress this multi-value image data, many methods exist.
With manuscripts output by digital photocopier or printer, it is often the
case that text area and photo area co-exist on a single sheet.
Additionally, with printer output images, there are many manuscripts that
mix both images created by computer, or so-called computer graphics (CG),
and photos and other such read-in images scanned from a scanner.
These CG and scanned images each have different image characteristics. For
instance, CG image areas include many flat areas in which the pixel values
fluctuate either uniformly or not at all. Furthermore, even within the CG
image area, the character area in which only the white and black binary
exists and the gradation area in which the original element value
drastically changes also exists. In contrast, in many instances the
scanned images area includes noise picked up during reading by a scanner,
causing minute fluctuations in the pixel value.
Additionally, the CG and scanned image boundaries have different image
characteristics. For this reason, effectively compressing mixed image data
with high quality requires the optimum compression process for each set of
image quality characteristics. In order to meet this need for image data
mixing small areas having varied image characteristics it is necessary to
select the optimum compression process for each area depending on the
image characteristics.
An adaptive image compression method has often been proposed. The CG area,
often including images requiring a high degree of resolution such as
characters, line drawings, and the like, a compression method in which the
resolution data does not deteriorate is preferred. For example, reversible
compression methods such as MMR, LZW, JBIG and the like and block
compression methods such as BTC and the like, in which the gradation data
deteriorates but the resolution data does not, are appropriate.
Scanned image areas often include images requiring gradation data more than
resolution data, for example photos, natural images, and the like, where a
compression method in which the gradation does not deteriorate is
preferred. In the case of applying the reversible compression method in
which the image does not deteriorate after decompression to the scanned
image area, the pixel values severely change in this area and entropy is
high, so it is not possible to effectively compress data by the reversible
compression method.
Therefore, the non-reversible compression method is applied for the scanned
images area. Among the non-reversible compression methods, a method which
is able to maintain the gradation data after decompression is used. For
instance, there is the Adaptive Discrete Cosine Transform (ADCT) method or
the like, typified by the JPEG baseline, which is used as the standard
encoding method for color facsimile.
One object of the invention is to select image compression means resulting
in a reduction in required memory in an image processing device.
Another object of the invention is to compare the compressed image data to
a target and reselect and recompress the data until the target is
satisfied.
Additional objects, advantages and novel features of the invention will be
set forth in part in the description which follows, and in part will
become apparent to those skilled in the art upon examination of the
following or may be learned by practice of the invention. The objects and
advantages of the invention may be realized and attained by means of the
instrumentalities and combinations pointed out in the appended claims.
SUMMARY OF THE INVENTION
The present invention has as its main purpose the reduction of the required
memory capacity using an adaptive image compression method in a digital
photocopier, printer, or the like. The memory capacity is set at less than
the original amount of data of the image to be compressed, so the image
compression circuit needs to compress the original image data in order for
the image data to fit into the memory. Where the compressed image data
exceeds the memory capacity, it is not possible to decompress the image
data completely, so it is necessary to set the image compression circuit
so as to enable the target encoding amount (target compression rate) to
fit into the memory and to compress the data in order to clear that
compression rate.
When using fixed-length data compression as a compression method, the rate
at which the image data is compressed remains fixed and relatively
constant no matter what the input image. If this fixed compression rate
clears the target compression rate, the encoding data amount will fit
within the reduced memory capacity regardless of the input image. In
general, however, with fixed length compression the methods in which the
image quality deterioration of the decompression image is not striking
contribute only marginally to memory capacity reduction because the
compression rate is approximately 1/2 to 1/4.
In contrast, in the case of the variable-length compression method, the
compression rate varies depending upon the complexity of the image data to
be compressed. Furthermore, the compression rate fluctuates depending also
on the parameter settings at time of compression, while at the same time,
the image quality of the decompression image fluctuates, depending upon
the parameter settings. Generally, in the case of the non-reversible
variable compression, when the parameters are set so as to increase the
compression rate, the image quality tends to deteriorate. Conversely, when
setting the parameters so as to improve the image quality of the
decompression image, the compression rate tends to deteriorate.
When using the variable compression method with extremely complex images,
the compression rate does not reach the target compression rate, and as a
result, it is possible to exceed the target encoding amount. In this
situation, it is necessary to degrade the image quality and set the
parameters so as to increase the compression rate in order to reach the
target compression rate.
When compressing mixed CG and scanned images using the adaptive image
compression method, which performs the appropriate compression process for
each image area, the compression rate fluctuates sharply. The fluctuation
rate depends upon both the proportion of each image area having different
image characteristics which is included in the input image, as well as the
image composition.
For example, when the entire image consists of complex scanned images, and
if the adaptive image compression is performed with the parameters set so
as to maintain image quality, then the compression rate of the entire
image becomes approximately 1/4-1/6. However, if the entire image is CG,
consisting of characters, drawings, and the like, with large areas of flat
background, then the compression rate becomes 1/100 or more even though
the compression parameters are set at the same settings as for the
aforementioned entire scanned images.
Because with conventional adaptive image compression the compression rate
fluctuates sharply depending upon the image composition, if the input
image needs to be less than the target encoding amount, then the minimum
compression rate needs to be set at approximately 1/4-1/6, for example, in
the case of wholly scanned images. However, it is impossible to
effectively reduce the memory capacity with this amount of compression.
If the parameters are set high in order to increase the compression rate of
the scanned images, the encoding amount can be held to the target encoding
amount. However, when compressing mixed CG scanned images, there is a
concern that there will be an overcompression which compresses the image
by a compression rate higher than necessary. This results in
greater-than-necessary deterioration in image quality when setting the
parameters for images in which the scanned images account for a small
proportion of the overall image.
When applying the adaptive image compression method to mixed co-scanned
images, there are cases which satisfy the target compression rate for the
image as a whole even though the compression rate of the scanned image
area is low. Thus, prior to the present invention, controlling the
compression rate of mixed images was extremely difficult.
The present invention was invented in order to solve the above types of
problems. In addition to minimizing deterioration of the decompression
image quality, by selecting the optimum compression process for each area
for image data mixing small areas having different image characteristics,
the invention also analyzes the image composition of the entire image data
and reflect the results of that analysis in the selection of the
compression process method of each area, providing an image processing
apparatus which is able to attain the target compression rate for the
entire image.
The image processing apparatus of the present invention has a selection
means which selects from among a plurality of compression means which
covers a certain area of the image data input by the input means. The
plurality of compression means consists of different compression methods.
Additionally, the apparatus has an analysis means which analyzes the image
characteristics of the entire image data and a selection means structured
so as to select the compression means depending upon the results of the
analysis performed by the analysis means.
Thus, image data from the input means is obtained and is analyzed by the
analysis means to determine what areas having which kinds of image
characteristics exist therein--in other words, the image composition of
the entire image data is analyzed and the compression parameters set for
each of a plurality of compression methods. During compression, the
optimum compression method is selected from among the plurality of the
compression methods as per the analysis results per area, and it is
possible to achieve the target compression rate for the entire image data
because the compression parameter value is used as an entire image data.
In another embodiment, the image processing apparatus is constructed so as
to enable the analysis means to calculate the proportion of areas having
different image characteristics included in the image data. Thus, by
setting the compression parameter to reflect the area ratio (which is
easily grasped by sight) it is possible to minimize the image
deterioration of the decompression image while achieving the target
compression rate.
A further embodiment of the image processing apparatus maintains the amount
of the encoding data as a target encoding amount which shows the
compression rate of the entire image targeted, and outputs the compression
rate as encoded data by the compression means which was selected by the
selection means. The image processing apparatus can monitor the encoding
amount of this encoding data by the encoding amount monitoring means, and
can compare the encoding amount with the target encoding amount by the
encoding amount comparison means.
In this manner, it is possible to achieve the target compression rate by
managing both the compression rate of the targeted image data and the
compression rate achieved by the compression means selected by using the
compression rate as the encoding data amount, and thus being able to
understand the status at any time.
In addition, the image processing apparatus is constructed so as to change
the analysis result and redo the selection of the compression means
whenever the encoding amount obtained from the encoding amount monitoring
means exceeds the target encoding amount, as compared by the encoding
amount comparison means. Thus, when the target encoding amount is exceeded
during compression by management of the encoding amount the compression
parameters are changed to reflect the image characteristics and the
compression method is reselected. Thus, it is possible to achieve the
target compression rate because of this feedback function.
In another embodiment, the image processing apparatus has an input means
which inputs the coded image data into page description language (PDL), a
rasterizing means for the rasterization of PDL image data, a plurality of
compression means consisting of different compression methods, a
compression means which covers a certain area of the raster data, and a
selection means which selects from the plurality of compression means.
Moreover, the image processing apparatus has a discernment means which
discerns the image attributes of the entire coded image data, and the
selection means is constructed so as to select the compression means
according to the discernment result.
When used for printers and the like, the image data described by the page
description language is received by the printer, and this described code
is interpreted and rendered into the raster image; and simultaneously, the
image composition and characteristics of the entire image are discerned
from the described code. Based upon this discernment result, compression
is performed by selecting for each area, from among the plurality of
compression methods, that compression means which should cover the raster
image data and the results stored in the code memory.
Additionally, the image processing apparatus is constructed so as to obtain
by the discernment means the area proportion of the area which is included
in the code image data having different image attributes. Therefore,
compression parameters reflecting area ratios easily grasped by sight are
set and it is possible to minimize the image quality deterioration of the
decompression while achieving the target compression rate.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram depicting the encoding circuit of the invention.
FIG. 2 is a block diagram depicting the compression mode switching circuit
of the invention.
FIG. 3 is a block diagram depicting the decoding circuit of the invention.
FIG. 4 is a diagram depicting one example of the encoding data format of
the preferred embodiment.
FIG. 5 is a block diagram depicting the encoding circuit of the second
embodiment.
FIG. 6 is a general diagram depicting the band raster of the second
embodiment.
FIG. 7 is a flow chart depicting the encoding process of the second
embodiment.
FIG. 8 is a block diagram depicting the encoding circuit of the third
embodiment.
FIG. 9 is a block diagram depicting the compression mode switching circuit
of the third embodiment.
FIG. 10 is a block diagram depicting the ADCT compression circuit of the
third embodiment.
FIG. 11 is a diagram depicting one example of the quantization table of the
third embodiment.
FIG. 12 is a block diagram depicting the decomposition circuit of the third
embodiment.
FIG. 13 is a block diagram depicting the ADCT decomposition circuit of the
third embodiment.
DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS
The following explains in detail the image processing apparatus of the
embodiments of the present invention by referring to the drawings. In the
following explanation, a monochrome image data consisting of an image of 8
bits/pixel is used as the original image data, However, the present
invention is not limited to this. The full color image of 24 bits/pixel of
RGB, LTaTbT, YCrCb, XYZ, Luv and the like, or the full color image total
of 32 bit/pixel of YMCK for each 8 bits/pixel are also applicable. The bit
number for per pixel can be either 8 or 16 bits; either is possible.
Figure one is a block figure which shows an example of the composition of
the encoded circuit of the image processing apparatus of the first
embodiment. A picture block size of 4.times.4 is used for this explanation
even though this embodiment is not limited to this. The image data, which
is input from the image input apparatus 1 acting as the input means, is
sent to the image analysis buffer 3 of the image analysis circuit 2 acting
as a temporary analysis means. The image composition analysis circuit 4
analyzes the construction included within the image by referring to the
data of the image analysis buffer 3.
The optimum compression parameter calculation circuit 5 refers to the
analysis result of the image composition analysis circuit 4 and determines
the threshold value which is used to determine the logic and therefore
which compression method is selected per block. This optimum compression
parameter calculation circuit 5 sets the threshold parameters so as to
keep the compression rate of the entire image below the target compression
rate.
For example, if the image is complex and there is a possibility of being
unable to achieve the compression rate which the image processing
apparatus targets, the apparatus sets the threshold value which most often
determines the compression mode which is able to achieve the higher
compression rate. In the case of a simple image, the apparatus sets the
threshold value which determines a compression mode in which there is a
low compression rate but a better image quality. As a result, it is
possible to bring the compression rate of the entire image closely to the
compression rate that the apparatus targets.
After the above image analysis process is completed, the image compression
process begins. In the raster block conversion circuit 6, the input image
is broken into 4.times.4 pixel block units and sent to the compression
mode switching circuit 7. The compression mode switching circuit 7 selects
which compression method is applicable among the plurality of the
compression methods, based upon the threshold value information obtained
from the optimum compression parameter calculation circuit 5. In this
embodiment, the compression mode switching circuit 7 is a selectable
compression method with four compression modes: block single-color
approximation compression mode, block run-length compression mode,
block-internal two-color approximation compression mode, and
block-internal four-color approximation compression mode.
The encoding data composition circuit 8 treats per-block encoding data as a
single unit and adds in front of that data a tag signal which shows the
selected compression circuit. In addition, it aligns the encoding data
which is output from the four compression circuits in one bit stream and
outputs it as encoding data.
The following explains each construction element of this embodiment in
detail.
The image input apparatus 1 is the interface which receives a raster image.
Just like a scanner, the image input apparatus 1 is considered to read the
manuscript and convert the image into digital data; and like an external
interface, the image input apparatus 1 receives the image from the outside
network directly as digital data, or the circuit which receives the raster
data from a decomposer which outputs the raster image data in the case of
a post script printer.
The raster block conversion circuit 6 is the circuit which outputs the
4.times.4 pixel as one unit as one block.
The image analysis circuit 2 consists of the image analysis buffer 3, the
image composition analysis circuit 4, and the optimum compression
parameter calculation circuit 5, and analyzes the image data received from
the image input apparatus 1. The image analysis circuit 2 receives one
page of image data from the image input apparatus 1 and then stores it in
the image analysis buffer 3.
By referring to the data in the image analysis buffer 3, the image
composition analysis circuit 4 analyzes the composition which is included
within the image. In other words, different areas included within the
image having different image characteristics--for example,
characters/drawings, CG, and scanned images areas--are discerned, the
coordinates which the area covers, the largest gradation number within the
area, and its degree of complexity are determined and the area proportion
per area is calculated.
In this embodiment, the image composition analysis circuit 4 calculates
each area proportion of (1) background area, (2) character/drawing area,
(3) CG area, and (4) scanned images area, and outputs that area proportion
to the optimum compression parameter calculation circuit 5. By referring
to the results from the image composition analysis circuit 4, the optimum
compression parameter calculation circuit 5 determines the threshold value
which is used in the logic which determines which compression method is
selected per block.
The minimum and maximum values of the coordinates of each area can be used
to calculate this area proportion for each area. In addition, a counter to
count the number of pixels included in each area can be installed and the
count value of this counter can be taken as the area. It is also
appropriate to install the counter, which breaks the input image into
blocks of a predetermined size and counts the number of blocks included in
each area, and the count of this new counter can be taken as the area.
The compression mode switching circuit 7, which acts as a selection means,
selects which compression method is suitable for each block among the
plurality of compression methods, based upon the threshold value
information obtained from the optimum compression parameter calculation
circuit 5. The details of the compression mode switching circuit 7 are
shown in FIG. 2. The compression mode switching circuit 7 is able to
select among these four modes: block-internal single-color approximation
compression mode, block run-length compression mode, block-internal
two-color approximation compression mode, and block-internal four-color
approximation compression mode.
The following is a detailed explanation of each compression method.
First, the block-internal single-color approximation compression mode is
the compression method which approximately expresses the entire block in
one color. It calculates the average of the pixel values within the entire
block, and expresses the entire block as an average value. In the case of
the 8 bit/pixel 4.times.4 pixel block, the original data amount can be
expressed below as:
8 bit/pixel.times.(4.times.4)=128 bit/block
The encoded data amount of the single-color approximation compression mode
within the block is only the 8 bit which shows the average value, so the
compression rate can be expressed as:
8/128=1/16
This compression mode applies to solid areas where a relatively
high-resolution expression is not needed, such as areas of uniform pixel
values like image backgrounds, etc., average uniform colors of thick lines
and CG graphics and the like.
The block run-length compression mode is the mode which approximately
expresses both the number of identical blocks that continue (run-length),
and the entire block, as one color. For such blocks, after performing the
block-internal single-color approximation process, the number of blocks
through which identical blocks continue is counted, encoded, and output as
continuous numbers and run length.
The encoded data amount of the block run-length compression mode combines
the 8 bit which shows the average value in the block and the 8 bit which
shows the run length, for 8+8=16 bit. The compression rate of this mode
fluctuates depending on the value which the run-length can take. As the
run-length becomes larger, the compression rate increases, and as the
run-length becomes smaller, the compression rate declines.
In case of the minimum value 2 of the run length, the compression rate is:
16/(128+128)=1/8
Thus, the compression rate of this block run-length compression mode
becomes 1/8 or more. This compression mode is applied to those areas which
do not need comparatively high-resolution expression, in which the
uniformed pixel value continues over a wide range such as the background
area of the image and the like.
The block-internal two-color approximation compression mode is the
compression method which approximately expresses the entire block with two
colors. The number of colors within the block is counted and when the
number of colors is less than two, the two colors becomes the
representative colors of the block. When the number of colors within the
block is three, the block is expressed by approximating the pixel value
within the block as two colors.
The method which approximately expresses using two colors for the pixel
value within the block is able to be adapted to existing limited color
techniques, such as the median cut method. The encoding data amount of the
block-internal two-color approximation compression mode consists of the
central value of the block .times.2 and the pixel flag which shows which
central value each pixel becomes.
The two central values is shown by each 8 bit and the pixel flag per pixel
can be indicated by 1 bit per pixel. In the case of the 4.times.4 block,
the data amount of the pixel flag is:
4.times.4.times.1 bit=16 bit.
Furthermore, the data amount of the central value is
8+8=16 bit
Therefore, the encoding data amount of the block-internal two-color
approximation compression mode is:
16+16=32 bit. Thus, the compression rate of this mode is 32/128=1/4.
This compression mode is applied to areas which require picture quality of
comparatively high resolution and which include the pixel value of two
colors within the block. It is applied, for example, to areas which
include the edges of characters/drawings and the like and areas which
include the dither matrix of the dither, etc., and CG gradations.
The block-internal four-color approximation compression mode is the
compression method which approximates the entire block using four colors.
The number of colors within such block is counted and the four colors
become the central value of the block where the number of colors is four
or less. Where there are five or more colors the pixel value of the block
is approximated and expressed using four colors. The method which
approximates and expresses the pixel value within the block using four
colors can be adapted to the established limited color techniques just
like the block-internal two-color approximation compression mode.
The encoding data amount of the four-color approximation mode within the
block consists of the central value .times.4 and the pixel flag which
shows which central value out of four will apply to each pixel. The four
central values are shown by each 8 bit and it is possible that the pixel
flag per pixel is shown by 2 bit. In the case of the 4.times.4 block, the
data amount of the pixel flag is:
4.times.4.times.2 bit=32 bit.
Moreover, the data value of the central value is
8.times.4=32 bit. Therefore, the encoding data amount of the block-internal
two-color approximation compression mode is
32+32=64 bit.
Thus, the compression rate of this mode is
64/128=1/2.
This compression mode applies to areas which need picture quality of
comparatively high resolution, which include many colors within the block.
For instance, it applies to areas which include scanned images and
complicated CG.
The encoding data composition circuit 8 (FIG. 1) takes the encoding data
per block as one unit and adds the tag signal which shows the compression
circuit which was selected before the encoding.
The addition of a tag signal per block is because the encoding data length
comprising one block differs with each compression mode. The bit number of
the tag signal, even at minimum, needs enough bits to show independently
the compression circuit. In this embodiment, because four circuits are
used for the compression circuit, it is good to have two bits or more for
the tag signal. The first embodiment uses 2 bits as the tag signal. FIG. 4
shows the format of the encoding data in each compression mode.
The compression mode switching circuit 7 receives the block image
consisting of the 4.times.4 pixel from the raster block conversion circuit
6 and calculates how many kinds of pixel values within the block are
available, that is, the number of colors within the block, using the
block-internal color-count circuit 21 (FIG. 2). The block-internal
color-count circuit 21 sends this color-count to comparator (1) 22.
Comparator (1) 22 compares the value calculated by the optimum compression
parameter calculation circuit 5, which is the threshold value 1, with the
color-count. As a result, if the number of colors is more than the
threshold, then the process continues on to processing by comparator (2)
23, and if the number of colors is less than the threshold, then the
process continues on to processing by comparator (3) 24.
Comparator (2) 23, just like comparator (1) 22, compares threshold value 2
which is calculated by the optimum compression parameter calculation
circuit 5 with the number of colors. As a result, if the number of colors
is more than the threshold value 2, then the block-internal four-color
approximation mode is chosen as the compression mode for this block. If
the number of colors is less than the threshold value 2, then the block
internal two-color approximation mode is chosen as the compression mode.
Comparator (3) 24, just like comparator (1) 22 and comparator (2) 23,
compares the optimum compression parameter calculation circuit 5 with the
colors. As a result, if the number of colors is more than the threshold
value 3, then the single-color approximation compression mode within the
block is chosen as the compression mode. If the number of colors is less
than the threshold value 3, then the block run-length compression mode is
chosen as the compression mode. Selection circuit 25 refers to those
determinations and switches the input block image to the appropriate
compression circuit.
FIG. 3 is a block figure which shows an example composition of the encoded
circuit of the image processing apparatus of this embodiment. The encoded
data which is output from the encoding data composition circuit 8 is sent
to the tag signal separation circuit 9 and a predetermined number of bits
of the tag signal is first separated from the peak data and isolated as
the tag signal.
Depending upon the content of this tag signal, it is possible to determine
which compression mode compresses the succeeding encoding data line.
Because the code length of one block differs depending upon the
compression mode selected, a one-block selection of the encoding data is
read out from the succeeding data depending upon this compression mode.
For example, from the encoding data 8 bit is read out in the block-internal
single-color approximation compression mode, 16 bit in the block
run-length compression mode, 32 bit in the block-internal two-color
approximation compression mode, and 64 bit in the block-internal
four-color approximation compression mode. The read-out encoding data is
sent to the input switching circuit 10.
Depending upon the tag signal, the input switching circuit 10 selects the
decompression circuit appropriate to the encoding data from the plurality
of decompression circuits 11 and sends the encoding data to the
decompression circuit. Depending upon the compression circuit, the
decompression circuit 11 has four decompression circuits: the block
single-color approximation decompression circuit, block run-length
decompression circuit, block-internal two-color approximation
decompression circuit, and block-internal four-color approximation
decompression circuit.
The block-internal single-color approximation decompression circuit
receives 8 bits of the encoding data, takes this to be the average value
within the block, paints the entire block with this average value, and
outputs the result.
The block run-length decompression circuit receives 16 bits of encoding
data and takes the first eight bits as the average value within the block
and interprets the last eight bits as the run-length. This decompression
circuit outputs continuously only the run-length number of blocks in which
the entire block has been painted with this average value.
The block-internal two-color approximation decompression circuit receives
32 bits of the data and considers the first 8 bits to be the first block
central value and the second 8 bits as the second block central value. It
then disassembles the remaining 16 bits one by one and matches each to one
block consisting of the 4.times.4 pixel=16 pixels to create a flag showing
the central value corresponding to the position of each pixel. The flag
per pixel is checked and if the flag is "0" the first central value for
the pixel of the position is applied. If the flag is "1" the second
central value is applied. This operation is carried out in one-block
segments for one block and outputs one block of image data.
The block-internal four-color approximation decompression circuit receives
64 bits of data and disassembles the lead bits into four segments of 8
bits each, or 32 bits in total into four central values. The remaining 32
bits are broken up into two-bit segments matching each 4.times.4 pixel
with two bits and creates a flag which shows the central value
corresponding to the position of each pixel.
The flag is checked at each pixel and if the flag is "00" the first central
value is applied to the pixel in that position. If the flag is "01", the
second central value is applied. If the flag is "11" and "12", then the
third and fourth central values are applied, respectively. This operation
is carried out in one-block segments and outputs one block of image data.
As described above, there is virtually no numeric value arithmetic process
in any decompression circuit, rather a simple logical operation is used.
Compared with the compression circuit, the decompression circuit has a
fairly simple construction. This makes image decompression with high speed
possible. This is because decompression from the encoding data to the
image data needs to be synchronized with the image output apparatus such
as the printer and the like, and it is possible to output the image with
high speed by simplifying the decompression circuit and gaining
decompression speed.
The decompressed single-block section of the image data goes through the
output switching circuit 12 which moves | | |
