 |
Description  |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an apparatus for combining a character
font in an image subjected to data compression by encoding and stored in a
recording medium.
2. Description of the Related Art
Image data prepared by converting an image obtained by a television camera
etc. to a signal which can be treated by a digital signal processing
apparatus such as a computer usually comprises a tremendous amount of
information. Accordingly, if this image data is recorded in a recording
medium as it is or is transmitted by modem by computer or the like, the
efficiency would be very bad. Therefore, conventionally, information of
image data has been compressed by encoding and then recorded in a
recording medium or transmitted.
This data compression is carried out as follows.
First, the image data is subjected to analog-to-digital (A/D) conversion
and stored in an image memory as digital image data. On the other hand,
where a character font is superimposed at a predetermined pixel position
of this image, the data of this character font is read out from a
character font memory and the data of the image memory at that pixel
position is substituted by the character font data for each pixel. In this
way, the image data on which the character font is superimposed is read
out from the image memory for every block of size consisting of 8.times.8
pixels, subjected to a discrete cosine conversion (hereinafter referred to
as a DCT conversion), and converted to a DCT coefficient. This DCT
coefficient is quantized, and this quantized DCT coefficient is subjected
to Huffman encoding and data compression and then recorded in the
recording medium.
Note that, Huffman encoding is an encoding procedure in which short code
words are assigned to symbols having a high generation probability based
on statistical analysis of the generation frequency of symbols of an
information source and, at the same time, long code words are assigned to
symbols having a low generation probability, whereby the average word
length of the code words is shortened.
The image data (coded data) recorded in the recording medium is subjected
to Huffman decoding and converted to the quantized DC coefficient and
further subjected to inverse quantization to become a DCT coefficient. An
inverse DCT conversion is carried out with respect to this inverse
quantization DCT coefficient for every block consisting of 8.times.8
pixels so as to restore the image data. This image data is once written in
the image memory, read out from this image memory, and converted to an
analog signal and output to a display device.
Where a new character font is added to image data (coded data) recorded in
the recording medium or where a character font which has already been
superimposed on the image data is substituted by a new character font,
after the original image data is restored from the encoded data,
processing for adding the new character data to this restored data must be
carried out. Namely, the encoded data read out from the recording medium
is subjected to processings such as Huffman decoding, inverse
quantization, inverse DCT conversion, etc., and then once written in this
image memory, and the character font is written in the image data written
in this image memory. Then, this image data is subjected to the
processings of DCT conversion, quantization, Huffman encoding, etc. and
recorded in the recording medium.
However, this type of decoding and encoding processing requires a long
time, and enormous power is consumed by a circuit performing these
processings.
SUMMARY OF THE INVENTION
Accordingly, the present invention has as its object to provide a character
font combining apparatus which can make the processing easier when
character data is added to image data after data compression and, at the
same time, can suppress the power consumption required for the processing
as much as possible.
According to the present invention, there is provided a character font
combining apparatus provided with image data memory means, decoding means,
character data storage means, encoding means, and substitution means.
The image data stored by the image data memory means is obtained by
dividing digital original image data into a plurality of blocks and
applying orthogonal conversion and encoding to the image data for every
block. The decoding means decodes the image data read out from the image
data memory means and finds an orthogonal conversion coefficient for every
block. The character data storage means stores the character font
constituted by at least the orthogonal conversion coefficient of one
block. The encoding means encodes the orthogonal conversion coefficient
for every block to obtain the encoded data and, at the same time, writes
this encoded data in the image data memory means. The substitution means
substitutes at least one block among the blocks of the orthogonal
conversion coefficients outputted from the decoding means by an orthogonal
conversion coefficient stored in the character data storage means and, at
the same time, outputs this substituted orthogonal conversion coefficient
to the encoding means.
Also, according to the present invention, there is provided a character
font combining apparatus provided with orthogonal conversion means,
encoding means, decoding means, character data storage means, and
substitution means.
The orthogonal conversion means divides digital original image data into a
plurality of blocks and finds the orthogonal conversion coefficient of the
image data for every block. The encoding means encodes the orthogonal
conversion coefficient and records the same in the memory. The decoding
means decodes the image data recorded in the memory and finds the
orthogonal conversion coefficient for every block. The character data
storage means stores the character font constituted by at least the
orthogonal conversion coefficient of one block. The substitution means
substitutes at least one block among the blocks of the orthogonal
conversion coefficients outputted from one of the orthogonal conversion
means and decoding means by the orthogonal conversion coefficient stored
in the character data storage means and, at the same time, outputs this
substituted orthogonal conversion coefficient to the encoding means.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram of a first embodiment of the present invention;
FIGS. 2(a)-2(d) are views showing an example of a DCT conversion and
quantization;
FIG. 3 is a view showing an example of the DCT coefficient of a character
font;
FIG. 4 is a view showing a group classification table of DC components used
for Huffman encoding;
FIG. 5 is a view showing code words expressing group numbers;
FIG. 6 is a view showing a zigzag scanning used when encoding AC components
among the DCT coefficients;
FIG. 7 is a view showing an example of finding the encoded data from the
quantized DCT coefficient;
FIGS. 8(a)-8(c) are views showing an example of decoding the image data by
an IDCT conversion and inverse quantization from the encoded data;
FIG. 9 is a view showing a method of finding a differential value of the DC
components in a substitution operation of the character font; and
FIG. 10 is a block diagram of a second embodiment of the present invention.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
The present invention will now be described with reference to embodiments
shown in the drawings.
FIG. 1 is a block diagram of an electronic camera to which a character font
combining apparatus of the first embodiment of the present invention is
applied.
Light coming from an object S photographed is focused by a focus lens 11.
The photographed image is focused on a light receiving surface of a CCD
(charge coupled device) 12. A large number of photoelectric conversion
elements are arranged on the light receiving surface of the CCD 12. Each
photoelectric conversion element corresponds to one pixel. The
photographed image is converted to an electrical signal by the
photoelectric conversion elements and inputted to an A/D converter 13.
The photographed image is converted from an analog to digital format for
each pixel at the A/D converter 13, subjected to the predetermined
processing by a not shown signal processing circuit, and then stored in an
image memory 14. When one frame's worth or one field's worth of image data
is stored in the image memory 14, this image data is divided into a
plurality of blocks by a memory controller 15 and outputted to a DCT
processing circuit 16 for each block. Each block is constituted by
8.times.8 pixels as shown in (a) of FIG. 2. The image data Pxy of this
block consisting of 8.times.8 pixels is subjected to two-dimensional DCT
conversion in the DCT processing circuit 16 and converted to a DCT
coefficient Suv as shown in FIG. 2(b). Namely, in the present embodiment,
the DCT conversion is adopted as the orthogonal conversion of the image
data.
The DCT coefficient Suv output from the DCT processing circuit 16 is
inputted to a quantization processing circuit 23 via a first switch 21 and
a second switch 22 for each block consisting of 8.times.8 pixels. The DCT
coefficient Suv is quantized using the predetermined quantization table
Quv as shown in FIG. 2(d) in the quantization processing circuit 23, and a
quantized DCT coefficient Ruv as shown in FIG. 2(c) is obtained.
Note that, the quantization table Quv of FIG. 2(d) is determined according
to for example a JPEG algorithm.
The quantized DCT coefficient Ruv outputted from the quantization
processing circuit 23 is H-coded at a Huffman encoding processing circuit
24 and substituted to the code words and then recorded in a recording
medium M as the encoded data.
Where a character is added to the original image, the DCT coefficient Suv
outputted from the DCT processing circuit 16 is not inputted to the
quantization processing circuit 23 for the block to which the character is
added, but the DCT coefficient Suv read out from a font ROM 17 and
corresponding to that character is inputted via the second switch 22 to
the quantization processing circuit 23.
This character added is inputted via a keyboard or the like connected to a
control circuit 20. The character code is outputted from the memory
controller 15 in accordance with this input operation, and data (DCT
coefficient) in accordance with this character code is outputted from the
font ROM 17. Namely, the DCT coefficient obtained by applying DCT
conversion to the character font of the block consisting of 8.times.8
pixels is stored in the font ROM 17. For example, in the case of a
character "5", as shown in FIG. 3, when the DCT conversion is applied to
image data CF1 corresponding to this character, a DCT coefficient CF2 is
obtained. DCT coefficients CF2 corresponding to various characters are
stored in the font ROM 17.
The switching operations of the first and second switches 21 and 22 are
controlled by the control circuit 20 provided with a computer. Namely, the
DCT coefficient Suv outputted from the DCT processing circuit 16 or the
font ROM 17 is alternately inputted to the quantization processing circuit
23.
A concrete example of the DCT conversion and quantization will be explained
using FIG. 2. The image data Pxy of the block consisting of 8.times.8
pixels is DCT-converted at the DCT processing circuit 16, whereby the DCT
coefficient Suv is obtained. In this drawing, S.sub.00 is a DC (Direct
Current) component, and S.sub.01 to S.sub.77 are AC (Alternative Current)
components. The coefficient S.sub.77 expresses the coefficient having the
highest spatial frequency.
The DCT coefficient Suv is calculated at the quantization processing
circuit 23 by the following equation using the quantization table Quv, and
the quantized DCT coefficient Ruv is obtained.
Ruv=round (Suv/Quv) {0.ltoreq.u, v.ltoreq.7}
The term "round" in this equation means approximating to the closest
integer. Namely, the quantized DCT coefficient Ruv as shown in FIG. 2(c)
is obtained by division and rounding off between the respective elements
of the DCT coefficients Suv and the respective elements of the
quantization table Quv.
Operation of Huffman encoding the quantized DCT coefficient Ruv will be
explained referring to FIG. 4 to FIG. 7.
The encoding method differs between the DC component R.sub.00 and AC
component (the quantized DCT coefficient Ruv other than the DC component
R.sub.00). The encoding of the DC component R.sub.00 is carried out as
follows.
First, a difference between the quantized DCT coefficient R.sub.00 of the
block to be encoded at present and the quantized DCT coefficient R.sub.00
of one preceding encoded block is found. It is decided which of the groups
shown in FIG. 4 this differential value belongs to. The code word
expressing the number of that group is found from the code table (coding
table of DC component) shown in FIG. 5. For example, when the quantized
DCT coefficient R.sub.00 of the block to be now encoded is 16 and the
quantized DCT coefficient R.sub.00 of one preceding encoded block is 25,
the differential value is -9, and therefore it is decided that the number
(SSSS) of the group to which the differential value=-9 belongs is "4" from
the group number table of FIG. 4. Further, it is decided that the code
word of that group number (SSSS) is "101" from the code table of FIG. 5.
Subsequently, the order of the differential value in that group in the
group number table of FIG. 4 is expressed by an additional bit. For
examples the differential value=-9 is seventh in order from the smallest
in the group of the group number (SSSS)=4, and therefore the additional
bit becomes "0110". Namely, the Huffman encode word of the quantization DC
component R(Y).sub.00 of the block which is now being encoded becomes
"1010110".
On the other hand, in the encoding of the AC component of the quantized DCT
coefficient, first, 63 quantized DCT coefficients are subjected to the
zigzag scanning in an order shown in FIG. 6 and rearranged to
one-dimensional array data. Then, it is decided whether or not the
respective quantized DCT coefficient values arranged in one dimension are
"0". When a quantized DCT coefficient is "0", that quantized DCT
coefficient which is "0" is counted. By this, a length of continuity of
"0"'s, that is, a run length (NNNN) is obtained.
Contrary to this, when the quantized DCT coefficient is not "0" the group
classification the same as that for the DC component is carried out and,
at the same time, the additional bit is obtained. The group classification
of the quantized DCT coefficient of the AC component differs from the
group classification of the DC component and is carried out for the
quantized DCT coefficient thereof per se. Namely, when the quantized DCT
coefficient is for example "4", a group number (SSSS) "3" is obtained
referring to a table in the same way as that of FIG. 4. Also, the
quantized DCT coefficient "4" exists at fifth place from the smallest in
the group of the group number (SSSS)=3, and therefore the additional bit
becomes "100".
Subsequently, the AC code table (not shown) is referred to, and for example
where the run length of data immediately before the quantized DCT
coefficient of "4" is "0", the code word "100" is obtained based on this
run length and the group number (SSSS)=3. Then, by combining this code
word "100" and the above-described additional bit "100", the
two-dimensional Huffman encoding word "100100" is obtained.
The result of performing the Huffman encoding for the quantized DCT
coefficient of FIG. 2(c) is indicated as the encoded data 100 of FIG. 7.
An explanation will now be made of decoding of the encoded data recorded in
the recording memory M referring to FIG. 1 and FIG. 8.
The encoded data read out from the recording medium M is decoded to the
quantized DCT coefficient Ruv (refer to FIG. 8(a)) of a block consisting
of 8.times.8 pixels in the Huffman decoding processing circuit 25. This
quantized DCT coefficient Ruv is subjected to an inverse quantization in
the inverse quantization processing circuit 26 using the quantization
table Quv (refer to FIG. 2(d)) used in the quantization processing circuit
23, whereby the DCT coefficient Suv (refer to FIG. 8(b)) is obtained for
each block consisting of 8.times.8 pixels. This DCT coefficient Suv is
inputted via a third switch 27 to the IDCT processing circuit 28. In the
IDCT processing circuit 28, inverse discrete cosine conversion is carried
out with respect to the DCT coefficient Suv, whereby the image data Pxy
(refer to FIG. 8(c)) is restored for each block consisting of 8.times.8
pixels.
This image data Pxy is written in the image memory 14 by the control of the
memory controller 15, and the original image array is restored in this
memory 14. In this way, when one frame's worth or one field's worth of
image data is restored and stored in the image memory 14, this image data
is read out from the memory 14 by the control of the memory controller 15,
converted to an analog signal at the D/A converter 29, and outputted to
the display 30.
Where it is intended to add a character on the displayed screen when the
encoded data recorded in the recording medium M is decoded and displayed
on the display 30, when the DCT coefficient Suv of the block corresponding
to the position at which that character is to be displayed is outputted
from the inverse quantization processing circuit 26, the third switch 27
is changed over to the font ROM 17 side (position indicated by a broken
line). Namely, the DCT coefficient Suv of the character font outputted
from the font ROM 17 is inputted via the third switch 27 to the IDCT
processing circuit 28 and subjected to the inverse discrete cosine
conversion, thereby being written in the image memory 14 as the character
data. Then, this character data is read out from the memory 14, converted
to an analog signal, and outputted to the display 30.
Note that, the character added on the displayed screen is inputted via a
keyboard or the like connected to the control circuit 20.
An explanation will be made of an operation for substituting a character
font to a different character for an image recorded in the recording
memory M in a state where the character font is added, and recording the
same again in the recording medium M.
The first switch 21 is changed over to an inverse conversion quantization
processing circuit 26 side (position indicated by a broken line), and the
second switch 22 is set to the first switch 21 side (position indicated by
a solid line). The encoded data of the image is read out from the
recording medium M, and the quantized DCT coefficient Ruv is obtained at
the Huffman decoding processing circuit 25. This quantized DCT coefficient
Ruv is subjected to inverse quantization at the inverse quantization
processing circuit 26, whereby the DCT coefficient Suv is obtained for
each block. This DCT coefficient Suv is inputted via the first and second
switches 21 and 22 to the quantization processing circuit 23.
The second switch 22 is usually connected to the first switch 21 side. When
the DCT coefficient Suv of the block for which the character should be
substituted is outputted from the inverse quantization unit 26, it is
changed over to the font ROM 17 side. By this, the DCT coefficient Suv of
the character font outputted from the font ROM 17 is inputted to the
quantization processing circuit 23, and the DCT coefficient Suv of that
block is outputted from the font ROM 17, substituted to the DCT
coefficient corresponding to a desired character, and inputted to the
quantization processing circuit 23.
In the quantization processing circuit 23, the DCT coefficient Suv is
quantized using the quantization table Quv. This quantized DCT coefficient
Suv is subjected to the Huffman encoding at the Huffman encoding
processing circuit 24 and recorded in the recording medium M as the
encoded data.
Note that, the substitution operation of the character font as described
above is the same also for a case where character data is newly added to
an image to which the character data is not added. Namely, when the
encoded data of this image is decoded to the DCT coefficient Suv by the
Huffman decoding processing circuit 25 and the inverse quantization
processing circuit 26 and quantized at the quantization processing circuit
23 again, it is substituted to the DCT coefficient of a desired character
output from the font ROM 17, whereby the character is added.
On the other hand, if the inverse quantization or quantization is always
carried out in the inverse quantization processing circuit 26 and the
quantization processing circuit 23 also for a block for which the
substitution of the character font is not carried out, the time required
for the substitution processing of the character font becomes long.
Therefore, the apparatus is configured so that a bypass switch 31 actuated
by the control circuit 20 is provided between the Huffman decoding
processing circuit 25 and the Huffman encoding processing circuit 24. The
quantized DCT coefficient decoded at the Huffman decoding processing
circuit 25 is inputted directly to the Huffman encoding processing circuit
24 via the bypass switch 31 for the block for which the substitution of
the character font is not carried out.
When the character font substitution operation occurs, and the Huffman
encoding circuit encodes the DC component, the Huffman encoding processing
circuit 24 performs a process different from new image data recording to
medium M as will be explained referring to FIG. 9.
In FIG. 9, the frames indicate the blocks F11 to F15 and F21 to F25 each
consisting of 8.times.8 pixels, respectively. Among the numerals in the
respective frames, the numerals on the upper side of the lines indicate
the DC components of the blocks, and the numerals on the lower side of the
lines indicate the differential values of the DC components, that is, the
differences between the quantized DCT coefficient R.sub.00 (DC component)
of the blocks and the quantized DCT coefficient R.sub.00 (DC component) of
preceding encoded blocks. For example, the DC component of the block F11
at the far left of the upper row is "13", and the DC component of the
second block F12 from the left of the upper row is "4". Accordingly, the
differential value of the block F12 is 4-13=-9. Note that, the blocks F11
to F15 of the upper row indicate the state before substitution of the
character font, and the blocks F21 to F25 of the lower row indicate the
state after substitution of the character font.
The image data are stored in the blocks F11 and F21 on the far left and the
blocks F15 and F25 on the far right, respectively. In the upper row, the
character data of "1", "2", and "4" are stored in the second to fourth
blocks F12 to F14, and in the lower stage, the character data of "3", "4",
and "8" are stored in the second to fourth blocks F22 to F24.
Here, it is assumed that the DC components of the character data of the
blocks F22 to F24 of the lower stage are "5", "2", and "7", respectively.
The data outputted from the quantization processing circuit 23 is the
differential value concerning the DC component, and the DC component per
se is not outputted. Namely, when the differential value of the block F22
is obtained, the DC component of the block F21 cannot be used. Therefore,
in the present embodiment, the differential value is obtained based on the
DC component "4" of the block F12 before the substitution of the character
data and the differential value "-9". Namely, the differential value of
the block F22 is a=5-(4-(-9))=-8, and (4-(-9)) corresponds to the DC
component of the block F21.
Note, such an operation in which the differential value is obtained is
performed, prior to the quantization, in the quantization processing
circuit 23.
The differential value b of the block F23 is obtained by the subtraction of
the DC component of the block F23 from the DC component of the block F22.
It is the same also for the differential value c of the block F24. Note
that, the DC component of the block F24 is changed by the substitution of
the character data, and therefore the differential value d of the block
F25 becomes a value different from that before the substitution of the
character data. Accordingly, it is necessary to calculate also this
differential value d in the same way as the block F23 and F24.
As described above, in the present embodiment, where a character is added
to the image data, DCT coefficients in the units of blocks consisting of
8.times.8 pixels among the DCT coefficients having the two-dimensional
array found by applying the DCT conversion to the original image are
substituted to the DCT coefficients of the desired character. Accordingly,
it is not necessary to completely restore the image information encoded
and recorded in the recording medium to write the character, and therefore
it is not necessary to perform the DCT conversion and IDCT conversion. For
this reason, the power consumption of the entire apparatus can be kept low
and, at the same time, a reduction of the processing time can be achieved.
Note that, in the present embodiment, where the character written in the
encoded data recorded in the recording medium M is substituted to another
character, the DCT coefficient Suv outputted from the inverse quantization
processing circuit 26 is inputted to the quantization processing circuit
23 via the first switch 22, but it is also possible to once store all of
the DCT coefficients Suv in the memory, overwrite the DCT coefficient Suv
of the character font on a predetermined DCT coefficient Suv in that
memory, read out all of the DCT coefficients Suv from that memory, and
input the same to the quantization processing circuit 23. In this case,
the overwriting of the DCT coefficient Suv of the character font is
carried out, whereby even in a case where the statistical nature of the
DCT coefficient Suv of the image as a whole is changed, the quantization
table Quv can be adjusted, and the optimum data compression by the
quantization processing circuit 23 can be carried out.
Note that, in the present embodiment, an addition or substitution of the
character data with respect to the image data was carried out in the form
of the DCT coefficient, but it is also possible to perform the same in the
form of the quantized DCT coefficient Ruv obtained by the quantization
processing circuit 23. Namely, it is also possible to set the quantized
DCT coefficient Ruv after performing the DCT conversion and quantization
as the character data to be stored in the font ROM 17, substitute the
quantized DCT coefficient Ruv of the image outputted from the quantization
processing circuit 23 by the quantized DCT coefficient Ruv from the font
ROM 17, and input the same to the Huffman encoding processing circuit 24.
Note that, in this case, the quantization table Quv used in the
quantization processing circuit 23 must be fixed.
Also, it is possible to adjust the density of the character in the restored
image by adjusting the value of the coefficient S.sub.00 (DC component)
existing in the block consisting of 8.times.8 pixels of the character
font. Therefore, in the processing of substitution by the DCT coefficient
of the character font, the value of the DC component in the DCT
coefficient Suv of the character to be substituted is increased or
decreased from an average of the values of the DC components of the
respective blocks existing in the vicinity of the block to be substituted
to perform adjustment.
Note that, the first to third switches 21, 22, and 27 can be any of an
analog switch, switches constituted by mechanical contact points, etc. and
are not particularly critical. Further, it is also possible to connect the
image memory 14 to the A/D converter 13, D/A converter 29, DCT processing
circuit 16, and the IDCT processing circuit 28 by a local data bus.
FIG. 10 is a block diagram of the character font combining apparatus
according to the second embodiment.
This embodiment shows a structure in which, a large volume of image data is
compressed and stored in a large scale memory device such as a storage 41
and an image data base is constructed. The character data is added to that
compressed image data. In the storage 41, a static image such as for
example a map image is subjected to the DCT conversion, quantization, and
encoding and stored as the encoded data.
The encoded data read out from the storage 41 is inputted to the Huffman
decoding processing circuit 25, and the quantized DCT coefficient Ruv is
found for each block consisting of 8.times.8 pixels. This quantized DCT
coefficient Ruv is inputted to the inverse quantization processing circuit
26, and the DCT coefficient Suv is obtained for each block consisting of
8.times.8 pixels using the predetermined quantization table Quv. This DC
coefficient Suv is inputted via the switch 42 to the quantization
processing circuit 23.
The switch 42 usually exists on the inverse quantization processing circuit
26 side (position indicated by the solid line). When the DCT coefficient
Suv of the block to which the character should be added is outputted from
the inverse quantization processing circuit 26, it is changed over to the
font ROM 17 side, and the DCT coefficient Suv of the block consisting of
8.times.8 pixels of the character font is inputted from the font ROM 17 to
the quantization processing circuit 23. The switching operation of the
switch 42 and the output of the character font from the font ROM 17 is
controlled by the controller 44.
In this way, among the DCT coefficients Suv outputted from the inverse
quantization processing circuit 26, the data at the desired pixel is
substituted to the DCT coefficient Suv of the character font outputted
from the font ROM 17 and inputted to the quantization processing circuit
23. This DCT coefficient Suv is converted to the quantized DCT coefficient
Ruv using the predetermined quantization table Quv in the quantization
processing circuit 23, encoded at the Huffman encoding processing circuit
24, and recorded in the storage 41.
Even in a case where the character data has already been added to the image
data stored in the storage 41 and that character data is to be substituted
to new character data, a similar processing to that described above is
carried out. Namely, the DCT coefficient Suv is decoded from the encoded
data read out from the storage 41 by the Huffman decoding processing
circuit 25 and inverse quantization processing circuit 26, and the DCT
coefficient Suv of the predetermined block is substituted by the DCT
coefficient Suv outputted from the font ROM 17.
Accordingly, also by the present embodiment, in the same way as the first
embodiment, the processing time of adding the character font to the image
data stored in the storage 41 can be shortened, and the power consumption
required for the processing can be kept low.
Also, in the same way as in the first embodiment, it is also possible to
provide the bypass switch 31 switched by the controller 44 between the
Huffman decoding processing circuit 25 and the Huffman encoding processing
circuit 24. By this, for the block for which the substitution of the
character font is not carried out, the quantized DCT coefficient decoded
at the Huffman decoding processing circuit 25 is input directly to the
Huffman encoding processing circuit 24 via the bypass switch 31, and the
time required for the substitution processing of the character font is
shortened.
It is also possible to perform the substitution processing by the DCT
coefficient Suv of the character font using the semiconductor memory in
the same way as the first embodiment. Namely, it is also possible to store
the DCT coefficient Suv outputted from the inverse quantization processing
circuit 26 in the semiconductor memory in exactly an amount of one pixel
of information, overwrite the DCT coefficient Suv of the character font of
the font ROM 17 on the above DCT coefficient Suv in the units of blocks,
read out the DCT coefficient Suv from that memory, and input the same to
the quantization processing circuit 23. In this case, the respective
coefficients of the quantization table used in the quantization processing
circuit 23 are adjusted considering the statistical nature of all DCT
coefficients Suv of the semiconductor memory for which the substitution of
the DCT coefficient of the character font is completed, and the quantized
DCT coefficient Ruv is obtained in the quantization processing circuit 23
using this adjusted quantization table. According to this structure, it is
possible to adjust the amount of information of the encoded data and
adjust the image quality of the restored image.
Also in the second embodiment, in the same way as in the first embodiment,
it is also possible to perform the substitution of the image data and the
character data not by the format of the DCT coefficient, but by the format
of the quantized DCT coefficient. Moreover, it is also possible to
increase or decrease the value of the DC component of the character font
considering the value of the DC component of the image data at the time of
substitution of the character data.
In both of the first and second embodiments, it is also possible to perform
the substitution in a state where the codes of the respective AC
components in the DCT coefficients Suv of the character font are inverted
considering the value of the respective DC components for which the
character data is added or substituted. By this, the dense and sparse
portion of the character font which is added or substituted can be
inverted, and the visibility of the character on the restored image can be
enhanced.
Note that, it is also possible to use other orthogonal conversions such as
a Fourier transformation, Hadamard's transformations Harr transformation,
etc. in place of the discrete cosine conversion.
Further, the size of the character font stored in the font ROM 17 was set
as 8.times.8 dots in the above-described embodiments, but it is also
possible to change this to the size of 16.times.16 dots etc. In this case,
four DCT coefficients Suv of blocks consisting of 8.times.8 pixels in
combination are stored in the font ROM 17 as one character, and image data
of four blocks are substituted by four DCT coefficients Suv of the
character font, whereby the character font is written in the image data.
Also, for characters having a size of 24.times.24 dots, nine blocks are put
together by the DCT coefficient Suv of the character font for
substitution.
In each embodiment, the size of the block of the conversion coefficient was
set to 8.times.8 pixels, but it is also possible to change the same to a
block size consisting of 4.times.4 pixels, 16.times.16 pixels, etc. In the
present invention, the block size thereof is not particularly critical.
However, the block size obtained by dividing the original image data and
one unit of block size constituting the character font stored in the font
| | |
