JPH046955A - Color picture coding system - Google Patents

Abstract

PURPOSE:To improve the coding efficiency and also the picture quality by eliminating specific noise caused to an area to be at a prescribed density in a binarized picture. CONSTITUTION:R, G, B input data 201, 202, 203 representing each picture element of a color picture are inputted to a surrounding state detection section 10. The surrounding state of a coded picture element is fed to a state prediction section 13 through lines 220, 221, 222 respectively. A state prediction section 13 decides the picture element state around the coded picture element and sends a state signal st206 and a coded picture element signal D to a coder 11. Moreover, when a noise pattern detection section 12 discriminates the coded picture element to be noise, the state prediction section 13 corrects the data of the coded picture element to encode a solid color. The noise pattern detection section 12 sends a solid signal 229 and the picture elements of R, G, B of the coded picture element corrected at that time to the state prediction section 13 through lines 226, 227, 228 respectively. A coder 11 generates a code 207, which is sent to a decoder 14.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a color image encoding method in a color facsimile device that communicates color images.

[Prior art]

The conventional image encoding method is G3. recommended by CCITT (International Telegraph and Telephone Consultative Committee). A run-length encoding method, typified by G4 facsimile, is generally used. This encoding method is a method that counts the length (run length) in which a pixel continues to be white or black, and determines a code corresponding to the count value from a code table prepared in advance. The code table used here is characterized by assigning relatively short codes to long white runs, which are common in document images.

On the other hand, recently, the development of inexpensive color printers has progressed, and color images, especially 1-bit red (R), green (G), blue (B), yellow (Y), cyan (C), and magenta (M)
Binary color image communication with different data has been proposed.

As a binary color encoding method, three colors are encoded for each bit plane, and MH for black and white is used.

A method using an MR encoding method has been considered.

[Problem that the invention is trying to solve]

However, with the above method of encoding the three colors for each bit plane, encoding each R, G, and B bit plane increases the entropy of the original R, G, and B information sources. , there is a problem that the encoding efficiency is poor.Simply put, this means that color correlation information is not used.

Therefore, the challenge is how to improve encoding efficiency by utilizing color correlation. In particular, a color binary image in which each color is pseudo-halftoned using a dither method, an error diffusion method, or the like is problematic because the color correlation is very low. Furthermore, parts other than halftone images,
For example, when the entire image containing constant density colors such as characters, multicolor colors, background, etc. is converted into pseudo-halftone, a unique noise pattern appears even in areas that should originally have constant density.

When such an image is encoded, there is a drawback that the efficiency of the portion where the encoding efficiency is originally high becomes extremely low.

The reason for this is as follows. When the R, G, and B components are independently binarized using the error diffusion method, etc., the parts where each component should be uniformly "0" or "l", that is, the part of the image that can be represented by eight colors (hereinafter referred to as (hereinafter referred to as the solid color part) contains a unique noise pattern due to error diffusion from the surroundings, and the surrounding "0"
This is because "1"s are sparsely scattered within "1"s, or "0s" are sparsely dotted around surrounding "1s".Some of the solid colors containing such noise have poor encoding efficiency. Therefore, it is necessary to create an image with noise removed. However, it is necessary to distinguish such noise from highlight parts or dark parts of the halftone image part.

〔Example〕

FIG. 1 shows an embodiment of a color image transmission system to which the present invention is applied. R, G, B representing each pixel of a color image
Each 1-bit input data 203202.203 is inputted to the surrounding state detection unit 10 from each frame memory (not shown). Reference numeral 10 is used to detect whether or not the surrounding area of a pixel to be encoded (encoded pixel) is a solid color. In particular, it is used to detect noise in a part of solid color in a binary color image displayed in pseudo-halftone.

Reference numeral 12 denotes a noise pattern detection unit, which detects whether the surrounding pixel state obtained by the surrounding state detection unit 10 of the encoded pixel is a noise pattern or not, and corrects the pixel value if the encoded pixel is noise. . The solid signal 229 is sent to the state prediction unit 13
send to

The surrounding pixel signals of each R, G, and B brane are 22 each.
3.224.225 to 12.

The surrounding state of the encoded pixel is detected by R, G from the surrounding state detection unit IO.
, B are sent to the state prediction unit 13 at 220, 221, and 222, respectively. The state prediction unit 13 determines the state of pixels around the encoded pixel and sends a state signal 5t206 to the encoder 11. The encoded pixel signal is also sent to the encoder 11.

Further, in the state prediction section 13, the noise pattern detection section 12
If the encoded pixel is determined to be noise, the data of the encoded pixel is corrected and a solid color is encoded. The noise pattern detection unit 12 sends the solid signal 229 and the R, G, and B pixel values of the corrected encoded pixels at 226.227.228 to the state prediction unit 13, respectively.

The encoded pixel signals are serially sent to the encoder in the order of RGB, RGB, . . . at 203.

In the encoder 11, a code 207 is generated based on the pixel signal and the state signal st by Markov model encoding such as an arithmetic code, as will be described later, and is transmitted to the decoder 14.

In the decoder 14, decoding processing is performed on the transmitted code 207, and the decoded RGB serial pixel signal D'
208 is obtained. D' 208 is a pixel signal R'G'B' separated and decoded by the RGB parallel conversion unit 15.
, 209.210.211 is obtained.

FIG. 4(a) is a detailed block diagram of the surrounding state detection section IO. Although FIG. 4(a) only describes the image signal R, G and B can also be represented by a similar block diagram. Image signal R200 is sent to line memories 10-1 to 10-5.
is temporarily stored. Line memories 10-1 to 10-5 temporarily store five lines around the encoded pixel.

a11+al□,..., a57 is 5× around the encoded pixel
This is a latch that sequentially stores the states of seven pixels. a 34 stores pixels corresponding to encoded pixels. The inner 3x5 pixel state (15 bits) of the 5x7 pixels is 223-
1 to the noise pattern detection section, the outer 20 pixels state is 2
Similarly, it is sent to the noise pattern detection section 12 at 23-2.

FIG. 4(b) shows surrounding 5×7 pixels including the encoded pixel. ○ indicates encoded pixels, ■ indicates inner 3×5 pixels,
The mouth is shown by 200 pixels on the outside.

FIG. 5(a) is a detailed block diagram of the noise pattern detection section 12. The surrounding pixel state of the encoded pixel (a34 in Figure 4) is the inner 3 x 5 pixels (155 pixels are 2 out of 5 x 7 pixels).
23-1 to lookup table (LUT) 12-3
is input to. The states of the outer 20 pixels among the 5×7 pixels are input to the AND section 12-1.12-2.

12-1 is 1 when all the outer 200 pixels are 1n.
12-2 outputs 1 to 12-5 when all are "0".

When the output 12-4 or 12-5 is 1, it is determined by the LUT 12-3 whether the inner 3×5 pixels are a noise pattern.

FIG. 5(b) shows an example of a noise pattern within 5×7 pixels. ■When there is a 1 pixel isolated point in the center,■,■
are noise patterns that appear in a binarized image using the error diffusion method. In the case of these patterns, the central pixel value is determined to be noise and is changed.

These patterns are stored in the LUT 12-3 in advance. LUT 12-3 detects whether or not the inner 3×5 pixels (155 pixels) are a noise pattern only when output 12-4 and output 12-5 are 1.

If it is a noise pattern, solid signal 229-1
is set to "1", and the changed encoded pixel value is output at 226. Although FIG. 5(a) shows only the image signal R, G and B can be constructed in the same way.

FIG. 9 is a detailed block diagram of the state prediction unit 13 (FIG. 1). Pixel signals R, G, and B are each 22o1221
．． 222 to the state prediction unit 13. The details of state prediction for G and B are the same as for R, so they will be omitted. 13-1 is 1
Line memory for storing the pixel state before the line, 131~
138 is a latch for storing surrounding pixels of the encoded pixel. The encoded pixel is the value stored in latch 136.

13-2 is a selector which selects whether the next encoded pixel value is the corrected pixel value from the noise pattern detection section. As described above, when the noise pattern detection unit 12 determines that the pixel is noise, 229-1 becomes "1" and by selecting the corrected pixel value 226, a solid color without noise can be encoded. Latches 131 to 138 are seven pixels surrounding the encoded pixel.

These are input to the selector 13-3 as a reference pixel value 13-4. Also, the modified encoded pixel value 13-5 is similarly input to the selector 13-3.

Similarly, in the case of G pixel value, the surrounding 7 pixels have a reference pixel value of 13.
-6 as encoded pixel 13-7 together with selector 13-
3, the R pixel value number is also output to the selector 13-3 as a reference pixel value 13-8 and an encoded image 13-9, respectively.

Selector 13-3 inputs R, G,
B encoded pixels and reference pixels serially R, G, B, R at 3x clock speed.

G, B, . . . are output to the encoder 11 (FIG. 1).

FIG. 10 is an explanatory diagram of reference pixels for determining the encoding state 5t206. (a) shows a reference pixel with an R pixel value number, and indicates that seven encoded pixels around the encoded pixel shown in cloudy color are referred to.

(b) shows reference pixels of the G pixel value number, and represents that eight pixels are referred to, including seven pixels similar to (a) and an R pixel at the same position as the encoded pixel.

Furthermore, FIG. 5(C) shows the reference pixels of the B pixel signal, and refers to 9 pixels including the same 7 pixels as in (a) and the pixels at the same position as the R1 and G encoded pixels. It represents.

In addition to referring to surrounding pixels of the same color in this way, the encoder 11
It is possible to increase the prediction matching rate and improve the coding efficiency. The selector 13-3 selects the reference pixels and encoded pixels shown in FIG. 10 in synchronization with the color instruction signal,
The encoded pixel signal 203 and encoding state 5t206 are output.

FIG. 2 is a block diagram of the encoder 11.

Before explaining FIG. 2, the arithmetic codes used in this embodiment will be explained.

As is conventionally known, arithmetic coding is a method in which a code is formed by arithmetic operations so that an input signal string becomes a code expressed in binary decimal numbers. This method uses Lang
The document “Comp
Ression of Black / White
eImages with Arithmetic
Coding, IEEETran Com,
Published in C0M-29, 6, (1981, 6), etc. According to this document, the already encoded input signal sequence is
, the probability of a less-favorable symbol (LPS) appearing is q, the arithmetic register Augend is A (S), and the code register is C (S).
Then, perform the following arithmetic operations for each input signal.

A(Sl)=A(S)Xq scream A(S)X2-Q...
(1) A(So) = <A(S)-A(Sl)>z
...(2) <>7 is a significant digit I! c(so) = C(S), which represents truncation in bits
...(3) C(Sl)=C(S)+A(So)
...(4) Here, if the encoded data is a dominant symbol (MPS: 0 in the above example), A (So), C
(So) is used to encode the next data. In addition, in the case of the inferior symbol (LPS: 1 in the above example), A (Sl),
C(Sl) is used to encode the next data.

The new value of A is multiplied by 28 (S is an integer greater than or equal to 0) and 0.
It falls within the range of 5<A<1.0. This processing corresponds to shifting the calculation register A eight times in hardware. The code register C is also shifted the same number of times, and the shifted out signal becomes the code. The above process is repeated to form a code.

Further, as shown in equation (1), the occurrence probability q of LPS is approximated by a power of 2 (2-Q: Q is a positive integer), thereby replacing the multiplication calculation with a shift calculation. In order to further improve this approximation, q is approximated by a polynomial of a power of 2, such as the expression (5). This approximation improves the worst point of efficiency.

(1”= 2-” + 2-02
...(5) Also, the arithmetic code has Q for each encoded data.
Since it is possible to switch the value of , the probability estimation section can be separated from the encoding.

In this embodiment, the probability is estimated while encoding as described above (a dynamic method is used).

A block diagram of the encoder 11 shown in FIG. 2 that performs the above arithmetic coding will be explained.

The state signal 5t206 from the state prediction circuit 13 is input to the counter memory 23 and the encoding condition memory 24.

The encoding condition memory 24 stores, for each state represented by the state signal 5t206, a dominant symbol MP5108 which is a symbol that is likely to appear, and an index 1107 indicating the encoding condition including the probability of appearance of LPS of an arithmetic code to be described later. has been done. The MPS 108 is input to the predictive conversion circuit 27, and the pixel signal D203 is input to the predictive conversion circuit 27.
A YN signal 101 is generated which becomes 0 when it matches S108. The YN signal 101 is input to the update circuit 25, and the update circuit 25 increments the count of the corresponding state among the count values stored in the counter memory 23 when the YN signal is O. And counter memory 23
If the count value 0106 stored in the count table ROM12 matches the set value MC105 from the count table ROM12, the index 1107 is updated in the direction in which it becomes thicker (in the direction in which the appearance probability q of LPS becomes smaller). (Inversion of MPS) Note that the count table ROM 22 supplies the update circuit 25 with the number MC105 of MPS shown in Table 1, which is determined corresponding to the index I representing the probability of appearance q of the LPS.

Furthermore, in the update circuit 25, the MPS 108 and the pixel signal D2
03 does not match, that is, Y from the prediction conversion circuit 27
When the N signal is 1, the index 1107 is updated in the direction in which it becomes smaller (in the direction in which the LPS appearance probability q becomes thicker). Also, when the index is 1 and a YN signal with a value of 0 comes, the MPS is inverted. (0→1 or 1→0).Output 1' 109 and MPS' 110 are the updated index values and are stored again in the encoding condition memory 24.

In the encoding parameter determination circuit 26, the index 11
The coding parameter Qll of the arithmetic code based on the value of 07
Set l in the arithmetic encoder 28. This arithmetic encoder 28
Now, the YN signal 101 from the predictive conversion circuit 27 is arithmetic encoded using the parameter Q111 to obtain a code 102.

Note that an initial value is given to the encoding condition memory 24, and I,
By not updating the MPS, static encoding can be easily achieved.

FIG. 8 is a block diagram of the predictive conversion circuit 27.

Pixel signal D203 and MPS108 are connected to EX-OR circuit 29
pixel signal D203 according to the logical formula in Table 2.
0 when and MPS108 match, 1 when they do not match
A YN signal 101 is output.

FIG. 3 is a block diagram of the update circuit 25. When the YN signal 101 is 0, the count value 0106 from the counter memory 23 is incremented by +1 by the adder 31 and becomes the signal C'l12. This value is compared 9 with MC105 from the count table ROM 22 in the comparator 33, and C'
If the value of MC matches the value of MC, update signal UPA11
Set 3 to 1. Furthermore, the YN signal 101 is an update signal U.
PB114, and UPA and UPB enter the index change circuit 35. Also, UPA and UPB are OR circuit 37
A logical OR is performed at , and the output signal 115 of the OR circuit 37 becomes a switching signal for the selector 32.

When the signal 115 is 1, the selector 32 selects the O signal 119 to set the value of the counter, and otherwise selects the output signal C'l12 of the adder 31 and outputs it as the counter update signal C'l16. This is stored in the counter memory 23
to be memorized.

The index change circuit 35 includes a signal dl17 (typically d=1) that controls the index update increments, and a UPA113. UPB114 and encoding condition memory 2
4 to the current index 1107 is input.

Table 3 is a table showing the index update method in the index change circuit 35 (Table 3 shows cases where the update increments are d=1 and d=2). , UPB to determine the updated index I'. Also, I=
1 and when UPB=1, EX signal 118 is set. E
When the X signal is 1, the inverter 36 inverts the symbol of the current MP310B (0→1 or 1→0) and updates the MP
Obtain S'110. Also, when the EX signal is 0, MP
S' is not changed. Updated I' 109 and M
PS' 110 is stored in the encoding condition memory 24 and used as index I and MPS for the next processing. Note that the updating method shown in Table 3 can be configured as a table using a ROM or the like, or can be configured as a logic using an adder/subtractor.

As described above, when MPSs as many as the number of MPSs determined according to the value of the index ■ representing the appearance probability q of an LPS approximated by a polynomial of a power of 2 occur, d is added to the index I, and the arithmetic code is The appearance probability q of the LPS used is made small, and on the other hand, when an LPS occurs,
The index I is subtracted by d, and the probability of occurrence q of LPS used for the arithmetic code is increased.Furthermore, if LPS occurs in a state where index I is 1, which indicates the probability of occurrence q of LPS is 0.5, Invert MPS.

In this way, the index and MPS are adaptively applied to the input image.
By updating , arithmetic coding with high coding efficiency can be achieved.

FIG. 6 is a coding efficiency curve of the arithmetic code used in this embodiment. Hereinafter, the value of index I will be indicated by a lowercase letter i. This curve is expressed by equation (6) when the LPS appearance probability is the approximate probability qei at the time of q1 encoding.

Then, the index I is sequentially set to 1. 2. 3. ...and given.

Here, the numerator is entropy, and qei is the value shown by formula (7).

qei=q 1 +q 2 −(7)
The value of qll q2 is the value of the polynomial approximation of a power of 2 and is the fifth
given in the table. For example, it is shown by (8) to (10).

qel = 2-'
...(8) qe2 = 2-'-2-'
−(9)qa3 =2−2+2−”
...(10), and the peak point qei at which the efficiency η becomes 1.0 in this probability will be hereinafter referred to as the effective probability. Furthermore, it is clear that efficiency is improved by calling the intersection of the efficiency curves the boundary probability qb+, and encoding using the effective probabilities adjacent to this probability.

In this embodiment, the effective probability qei shown in Table 4 is selected from the probabilities that can be approximated by two terms as shown in equation (5). Furthermore, Q II Q2+ Q3 in Table 4 is a parameter Qc111 sent to the arithmetic encoder 28. That is,
Ql and Q2 are the shift amounts given to the shift register,
This shift operation performs a power of 2 calculation. Further, Q3 indicates the coefficient of the second term, and performs switching between 10 and -.

The values of MC in Table 1 are determined as follows.

That is, when the number of LPS is NL and the number of MPS is NM,
The probability of LPS occurrence is given by equation (11).

Solving this equation using NM gives equation (12).

NM= LNL (1/(11)J...(
12) However, LXJ represents rounding up to the nearest whole number.

By giving qbt shown in FIG. 6 to q in equation (12), the number of dominant symbols (MPS) there
Mi is calculated. Therefore, MC is calculated from equation (13).

MCi =NM+++ −NM+
... (13) The value of MC in Table 1 is expressed by formula (11),
It is calculated from (12) and (13) with Nt = 2.

In this way, the number NMi of dominant symbols MPS corresponding to each index I is determined based on each boundary probability qbi as shown in FIG. shall be.

The value of this MC and the number of dominant symbols generated are compared as described above, and if the value of MC and the number of dominant symbols match, the state is suitable for encoding using the adjacent index. It is determined that index I is changed. As a result, the index (2) is changed at a good timing based on the number of occurrences of dominant symbols, and encoding using the optimum index I can be adaptively achieved.

FIG. 7 is a block diagram of the arithmetic encoder 28.

Of the control signals Qcl 11 (Table 4) determined by the code parameter determination circuit 26, shift register A
Ql to 70, Q2 to shift register B, selector 72
Q3 is input to . Ql and Q2 are the Augend signals As12 for shift registers A and B, respectively.
Indicates how many bits to shift 3 to the right.

The shifted results become output signals 130 and 131.

The signal 131 is complemented by the inverter 76, and the selector 72 selects the signal 131 or the output signal of the inverter 76 using the control signal Q3 to obtain the output signal 132.

Adder 73 receives signal 130 from shift register A70.
and the signal 132 from the selector 72 are added, and AS
An I signal 124 is output. The subtracter 74 subtracts the As+ signal 124 from the As signal 123 to obtain As.

A signal 125 is obtained. The selector 75 receives the Aso signal 125.
and the As+ signal 124 are selected by the YN signal 101. That is, when the YN signal is 1, the Aso signal becomes the A' signal 126, and when the YN signal is O, the AS+ signal becomes the A' signal 126. In the shift circuit 80, processing is performed to shift the A' signal to the left until the MSB becomes 1, and this shift causes As
' signal 127 is obtained. The shift signal 132 corresponding to the number of shifts enters the code register 79, and from the code register 79, bits corresponding to the number of shifts are sent.
are output in order from MSB to become code data 130.

The code data 130 is processed using a bit processing method (not shown).
The bitl sequence is processed so that it is within a finite number of bits, and is transmitted to the decoder 14 side.

Further, the content CR128 of the code register 79 is stored in the adder 7
7, it is added to the Aso signal 125 and enters the selector 78. Also, the signal CR to which the Aso signal 125 is not added is
128 also enters the selector L/actuator 78, when the YN signal 101 is 1, CR' = CR, and when the YN signal is 0, CR' = C
It is output as a CR' signal 129 which becomes R+Aso.

The shift processing described above regarding the code register 79 is CR
'Do this for the signal.

In the embodiments described above, when encoding a binary R, G, B color image, the efficiency is improved by removing noise peculiar to binary processing such as error diffusion. It is also possible to perform noise removal and encoding separately. That is, prior to encoding the entire R, G, and B image planes, the encoding can be performed after performing noise removal processing on each of them.

As described above, when predictively encoding a color image signal represented by multiple color component signals, in addition to referring to surrounding pixels of the same color when predicting a coded pixel, other encoding methods are already used. By referring to color pixels, prediction efficiency is improved and coding efficiency is improved. In addition, the encoding efficiency is improved by detecting the partial noise of the pseudo-halftone solid color, which is the problem mentioned above, and converting it to a noise-free solid color before encoding. At the same time, some noise in the solid color was corrected,
It has also become possible to improve image quality.

No. table (-) is don' c　a　r　e.

〔Effect of the invention〕

As explained above, in the present invention, two
By removing the unique noise that occurs in areas that should have a constant density in the converted image, it is possible to increase the encoding efficiency and improve the image quality at the same time.

[Brief explanation of drawings]

Fig. 1 is a block diagram of a color image transmission system to which the present invention is applied, Fig. 2 is a block diagram of an encoder, Fig. 3 is a block diagram of an update circuit, and Fig. 4(a) is a block diagram of an ambient condition detection section. Block diagram, Figure 4(b) is a diagram showing 5x7 pixels around encoding, Figure 5(a)
is a block diagram of the noise pattern detection unit, FIG. 5(b) is a diagram showing an example of a noise pattern, FIG. 6 is a diagram showing a coding efficiency curve, FIG. 7 is a block diagram of the arithmetic encoder, and FIG. 9 is a block diagram of the prediction conversion circuit, FIG. 9 is a block diagram of the state prediction unit, and FIG. 1O is a reference pixel explanatory diagram. 10... Surrounding state detection section 11... Encoder 12... Noise pattern detection section 13... State prediction section Fig. 3 Intenix umbrella wheel ■ Fig. 5 (shi) ■ ■

Claims (1)

[Claims] A binary color image encoding system represented by a plurality of color component signals, characterized by comprising means for detecting noise existing in a constant density region and means for changing the pixel value of the noise. method.