JPS6012865A - Highly efficient encoding system of picture data - Google Patents

Abstract

PURPOSE:To encode various matrixes in an extremely wide range by the same method by encoding these matrixes on the basis of the periodicity of the matrixes arranged two-dimensionally. CONSTITUTION:Input binary signals Y(2), M(2), C(2), Bl(2) are accumulated in a line memory 33, an address signal from a reference point ROM35 is added to an output from the ROM35 or a moving vector signal IV obtained by the transfer of a control signal by an adder 34 and the contents of the memory 33 are read out by using the added signal as an address signal. A coincidence deciding circuit 37 decides the coincidence between a forecasting signal read out from a forecasting ROM36 by using the read out contents as an address signal and the input binary signal and outputs forecasting error signals Y(e), M(e), C(e), Bl(e) as the binary signals of the coincidence or inconsistency. At the time of binary encoding, a periodically arranged dither matrix is used as a threshold and a reference picture element is determined in accordance with the periodical arrangement of the dither matrix.

Description

DETAILED DESCRIPTION OF THE INVENTION (Technical Field) The present invention relates to a highly efficient encoding method for image data, and particularly to a method for encoding [8 data] that is binarized using a dither matrix.

(Background Art) A method of converting image density data into an a value using a dither matrix is conventionally known. Predictive encoding is also known as a method of encoding the a-valued data.

In addition, we have developed a method for producing an output image with halftone dots using a dot-concentrated dither matrix, and a method for producing an output image with halftone dots in the case of color image output, and by giving a screen angle to the halftone dots in the case of color image output. Sub-methods have also been proposed that eliminate visual obstacles such as:

As mentioned above, the encoding method for image data having a halftone structure with a screen angle has the drawback that the processing is too complicated in the encoding methods proposed in the past. In particular, the above drawbacks are significant when a different screen angle is provided for each color, such as in a color image.

(G objective) Therefore, the purpose of the present invention is to eliminate the above-mentioned drawbacks of the conventional method, and at the same time, to encode image data processed using dither matrices other than the dot concentration type with high efficiency using the same method. The object of the present invention is to provide a highly efficient encoding method for image data.

(Example) The present invention will be described in detail below with reference to the drawings. First, in order to generate halftone dots, for example, the first [2,! Co-value conversion is performed using a dither matrix DbiB with increased dot concentration as shown in (4). In addition, in order to simply give the screen angle, a dither matrix D is used as shown in Figure 1 CB).
A two-dimensional arrangement of M is determined in the main scanning direction and the sub-scanning direction.

Here, k indicates that one dither matrix DMK is moved by a dot in the sub-scanning direction in order to give a screen angle.

FIG. 2 shows an example of a circuit that performs a-value conversion using such a threshold matrix DM. Here, the input data ID representing gradation operation information usually consists of time-series pixel data of 6 to g bits, and is compared with each component of the threshold matrix circuit MTX in the comparator COM, and is of l bit representing white or black. It is output as output data OD.

As a result, if the threshold matrix DM is of the above-mentioned dot concentration type, a halftone image can be obtained, and if the threshold matrix DM is of the above-mentioned dot concentration type (IJ) configuration, a desired screen angle can be obtained. I can do it.

When converting the input image into a signal, the dot density is 3 per In.
.about.10 points, in such a high-resolution WJ image signal, the change in signal between adjacent pixels is small. Therefore, when performing co-value conversion using a dot-concentrated dither matrix DM as shown in FIG. 1, the black portions are very often points that spread out from the center depending on the signal level. Therefore, the co-valued signal forms halftone dots, the interval between the halftone dots is determined by the size of the dither matrix, and the screen angle is determined by the one-dimensional arrangement of the dither matrix.

In other words, the information regarding the halftone dots such as the halftone dot spacing of the data obtained by the above-mentioned U-value conversion has little dependence on the input drawing similarity, and the input/output system gutter as part of the encoding, especially the (Dezama) Much depends on the one-dimensional arrangement of the IJ. Therefore, adaptive processing for the input image, such as changing the reference point depending on the characteristics of the input image, or performing preprocessing (for example, replacing image signals or changing the encoding method depending on the type of image) before normal encoding, is necessary. 2 above without using
Highly efficient encoding can be easily performed using dimensional arrangement.

Here, the dither matrix used for the above-mentioned co-value conversion is I
xII array. Furthermore, if the matrix arrangement is moved by a dot in the sub-scanning direction as shown in FIG. xyy), point AO'
Good prediction can be made by using (X+Cy-k) as a reference point during prediction.

Here, regarding reference points, reference pixels (hereinafter referred to as nearby reference points) when determining a pixel to be used as a reference for a pixel to be pre-t111 within ° size IxI pixels of the dither matrix, and A pixel in which the above reference pixel is extended outside the ItX1 pixel range due to periodicity (hereinafter referred to as a periodic reference point), and a pixel that is a corresponding reference point to the predicted pixel itself due to the periodicity of the dither matrix (hereinafter referred to as a corresponding reference point) ) are collectively referred to as the total reference point.

Furthermore, as an extension of the above, the predicted pixel
Determine the reference point in the vicinity of .

Referring to FIG. 7, adjacent reference points AI, , A2p A3 . "" e When AL= (xijyl) for AN, point A4"4 (X4 + ly "It k) (
1) (Aly A2*...e ANt A
O: A1'l A2't..., AN') Good prediction is made by using 2 total reference points.

As an example, a dither matrix DM' having a shape as shown in FIG.
A case will be explained in which the following are arranged as shown in FIG. Here, we will use a moving vector (klyk2) instead of a moving scene with a matrix size of 11. As shown in FIG. 7, the predicted pixel is AOs, and the nearby reference pixels are Aly A2p A3. At=(Xi+
71), A1"-(XL+ktvyt-A2) (/'
), then all reference points are (Ale A2* A3s AoF, At', A2',
A3') Good compression ratio can be obtained by encoding the signal.

A specific example will be given below for explanation.

If the dither matrix DM' shown in FIG. 9 is used in the arrangement shown in FIG. 9, the movement vector will be (+W=Z). Also, if the coordinates of Ao are (xotyo), its neighboring reference points AI, A2y A3 are Alo(xo -/+ yo
) A2 = (xoy yo −/) A3 = (XO+ /, yg −/). formula(
l'), if we add the reference point AC1'y A1'+ A2'+ A3' obtained from the two-dimensional distribution of the matrix, we get
Ao' = (xo + 11 yo -, 2) AI'
= Cxt) + 3r yO-2) A2' = (xo
+lI, yO-3) A3' = (xo + tr yo').

All reference points obtained from this (Ale A2t A3t
AO'y Al'yA21. A3/) is Fig. 7 ω)
become that way. Using this reference point, the value of A is determined as 1(t
I do it. The prediction signal So is statistically determined from a lot of image data. Mma-b.

P(Ao IAx, A2s Aa, A□', Al
o, Ai, A3') is the Token+1 probability that AQ becomes black with the state (black and white) of the reference point as a variable, then the predicted signal So I' is determined by the following.

Now P point (Azy A2F Aa, Ao', AI
', A2', A3') is 71 for black and 0 for white, and the form ff1i('t 'y 01 /e '* '* ('
), then P(AOI /, i, 0./, t, t, o
) has been statistically determined in advance. Here, P (
Aol /, t, o, /, /, /, o'
Let's say j=o, g.

From the above, it can be seen that the prediction signal SQ=/.

In reality, a fixed memory (ROM) is prepared in which the seven signals up to %Al''A3' are used as addresses and the prediction signal So determined as described above is stored.Therefore, the SROM address //θ By reading //10 (, l), the predicted signal SQ=/ is obtained.Here, 1If the signal at the actual point AQ is l, the error signal E=0, which is also 0.
Then, the error signal E=/, and this error signal E is taken as the signal for point AQ. Furthermore, highly efficient encoding of the signal string consisting of this error signal E can be achieved by using depth encoding or MH code.

When handling color images, it is necessary to change the screen angle for each color, so the reference point must be changed for each color, but in the present invention, the reference point for each color can be changed by changing the parameter kbk2. can be easily obtained.

FIG. 3 is an explanatory diagram of a color image recording apparatus to which the present invention can be applied. The color image recording device shown in FIG. 3 outputs uniform color information using an electronic copying device (laser beam printer) that includes a plurality of photosensitive drums arranged side by side, and sequentially prints color images formed by this electronic copying device. This is a device that records images in different colors overlappingly.

In FIG. 3, /ax/d is a scanning optical system, which extracts necessary image information from an image memory, etc. (not shown) as a light beam (laser beam), and this light beam is cyan (C), The image is formed on photosensitive drums UaP-λd arranged in parallel corresponding to magenta (Foundation), yellow (I), and black (BIJ). A developing device 3 is located near the photosensitive drums 2a to 2d.
a to 3d are arranged, and each photosensitive drum 2m-3 is arranged on the conveyor belt 7 side for conveying recording paper (not shown).
, 2d are provided with chargers IIa to Gd.

To explain the operation of the color image recording apparatus having such a configuration, the modulated light beam output from the scanning optical system /a''/d forms an optical image on each photosensitive drum 2aN2d, and then an electronic Through the photographic process, this formed image becomes an electrostatic latent image, which is developed by developing devices ja to 3d, and each color is sequentially transferred to the recording paper held on the conveying belt 5 by charging devices a to lId. A color image is formed using the scanning optical system /a to /d shown in Figure 3.
FIG. 2 is a schematic perspective view showing a detailed example of one of them (generally referred to by the symbol l), in which a light beam modulated by a semiconductor laser // is collimated by a collimating lens 10 and a rotating polygon mirror /2; The light is deflected by the The polarized light beam is passed through an imaging lens called an fO lens.
The photosensitive drum 2 (the above-mentioned photosensitive drum, 2aN, 2-
d (generally denoted by the symbol λ)) and beam scanning is performed. During this beam scanning, the tip of the 7-line scan of the light beam is reflected by mirror /q, and the photoelectric detector /
! Guide light to r. The detection signal from the photoelectric detector/3 is used as a well-known synchronization signal in the scanning direction H (horizontal direction). This signal name will hereinafter be referred to as a BD double signal or a horizontal synchronization signal.

FIG. 1O is an overall block diagram showing the signal processing system of the present invention.

In FIG. 10, ko/ is an input device, and the color image information of blue (B), green expansion), and red @) outputted from this input device is, for example, g bits (-5
6 levels) to represent gradations. These g-bit B, G, and R outputs are supplied to a masking/base-in circuit Ω to convert them into g-bit configured yellow (Y), magenta αυ, cyan (C), and black (BJ) signals, and then A predetermined /bit λ value conversion signal Y is supplied to the λ value conversion circuit 23.

Convert to M(,2), C(2) and BJ(ko), respectively.

These λ-valued signals are supplied to the compression encoding circuit, 2G,
/ Compressed signal Y (r) of less than 1 bit per image purple.

M (c) form v C (c) and nz (c),
These compressed signals are stored in the storage device J. From the storage device 2, a compressed signal Y (
c).

Read M(c), C(c) and BA'(c). The read and compressed signal is sent to a decoding circuit,
2 B, compression encoding circuit, λ value signal Y before 21I
(-2L M (,2L C(,2) and BJ(,
2) and then taken out to the outside by an output device: 1.7.

The compression encoding circuit described above is configured as shown in FIG. 1, for example, and the λ value signal Y(,2) from the binarization circuit 23
M(λ), C(2) and BA! (,2) is converted into a prediction error signal Y<@)e M(eL C
(e) and n7! (e) and convert these prediction error signals [run-length encoding circuit or MH encoding circuit 3
The compressed signal Y (c) * M (c) −
Obtain C(c) and Bl(c) O Further, the above-mentioned predictive encoding circuit 3/ can be configured as shown in FIG. 72, for example. In Figure 7.2, the input λ value signals Y (2) , M (2) t C(
λ) and BJ (ko) are stored in the line memory 33, and the address signal from the reference point ROM 33 and the % ROM
An adder 34' adds movement vector signals IV output from (not shown) or diverted from a control signal, and the contents of the line memory 33 are read out using the addition result signal as an address signal. The prediction signal read from the prediction ROM 36 and the input λ value signals Y(,2)2M(2), C(,2) and Bl! using the read contents as an address signal are read from the prediction ROM 36. (, 2) is determined by the coincidence judgment circuit 37, and as a signal of the λ value of the coincidence or mismatch,
Prediction error signal Y (eL M(e), C(e) and BJ
Output (e).

In the present invention, since the address signal for the line memory 33 is obtained as a sum signal of the output signal from the adder 3-hiro to the reference point ROM 3 and the movement vector signal, it is possible to change the arrangement of the dither matrix or the matrix itself. It is possible to encode changes easily and efficiently.

In the present invention, as a further extension of the above example, a plurality of dither matrices, for example A = (ag)
* For B = (btj), if you use a smaller value ε compared to the difference in darkness values in each matrix, and can express it as aij ”” bij+ε (ko), without distinguishing between %A and B. Movement vector (
kl+2) allows the reference point to be determined with the method of the invention.

Examples of dither matrices A and B in this case are shown in FIG. 1J(A) and FIG. 13), respectively.

Here, ε=l. As another example, if we consider the case where the dither matrix DM' shown in FIG. 11I is used, this is the small matrix A shown in FIGS. 15(4) to 15(B) (2).

B, C and D are two-dimensional movement vectors (,2,O)
It can be seen that the equation (2) is satisfied.

From this, an error signal sequence can be obtained by performing prediction using one reference point using a method similar to the encoding method of the above-described specific example. This method can be highly efficiently encoded using MH encoding, run-length encoding, etc. Figure 16 shows two rows of λ value conversion using the dither matrix DM' in Figure 1. As is clear from the figure, small matrices A, B, C
, D are arranged in a manner that is repeated periodically with a movement vector ('t o) such as A→B→A→B or C→D→C→D. That is, here, predictive encoding is performed using a pixel one pixel before in the main scanning direction as a reference pixel (corresponding pixel). In other words, the point immediately before the predicted dot map in the main scanning direction is used as the reference point.

In addition, when block encoding is performed for each data corresponding to a small matrix as described above, the present invention provides a good compression rate by performing encoding that takes into consideration the arrangement of the matrix and its periodicity. be able to.

As explained above, since the present invention performs encoding based on the periodicity of the matrix, it can be effectively applied to any dither matrix other than a dot-concentrated matrix if it is arranged with periodicity. I understand that there is something.

(Effects) According to the present invention, it is possible to easily obtain reference points for making predictions for dot-concentrated dither matrices, and it can also be applied to dither matrices that are not square matrices. Good encoding can be performed using a simpler method. Furthermore, when using multiple dither matrices, the parameter (kb kz)
Since it can be encoded in the same way by
This is particularly advantageous when a different matrix is used for each color in a color image, or when a matrix suitable for the nature of the input image is used.

Moreover, since the present invention performs encoding using the periodicity of the λ-dimensional arrangement of the matrix, by expanding it to the case where all the threshold values of the applied P matrix satisfy equation (2), a very wide variety of The same effect can be achieved by applying the same method to a matrix.

[Brief explanation of the drawing]

Fig. 1 (4) is a diagram showing an example of a dot concentration type dither matrix, Fig. 1 (ω) is a diagram showing an example of a two-dimensional arrangement of the matrix, and Fig. 2 is a block diagram showing an example of a λ value conversion circuit. , fA3 is an explanatory diagram of reference points in the present invention, FIG. 7 is an explanatory diagram of nearby reference points in the present invention, FIG. 3 is a diagram showing an example of a dither matrix that is not a square matrix, and FIG. 6 is a diagram of the matrix in the present invention. A diagram showing an example of a one-dimensional arrangement of
FIG. 71 is an explanatory diagram of nearby reference points in the present invention, FIG. 7(B) is an explanatory diagram of all reference points in the present invention, and FIG. 3 is a diagram schematically showing the color image recording device according to the present invention. 9th
The figure is a perspective view showing the schematic configuration 2 of the scanning optical system, and the 1st θ figure is a block diagram showing an example of the signal processing system.
/The figure is a block diagram showing a detailed example of the compression encoding circuit.
Figure 1: Figure 1 is a block diagram showing a detailed example of the predictive encoding circuit; Figures 13 and 73 are diagrams showing multiple dither matrices A and B; Figure 1 is a block diagram showing a detailed example of the predictive encoding circuit; 73(A) to 13) are diagrams illustrating small matrices that equalize the dither matrix shown in Figure 1, and FIG. 16 is an explanatory diagram of encoding according to the present invention. pg' in D; DM'... dither matrix, ID.
...Input data, MTX...WJ value matrix U path, COM...Comparator, OD...Output data, /ax/d...Scanning optical system, 2a N2d...Photosensitive drum, 33~ 3d...Developer, lI&~Hirod...Charger, Ri...Transport belt, 〆0...Collimating lens, //...Semiconductor laser/Co...Rotating polygon mirror, /3 ...Imaging lens, /4'...Mirror, /Left...Photoelectric detector, -?...Input device 1. 22...Masking/M input circuit 1.23...λ
Value converting circuit, 2G...compression encoding circuit, J...storage device 1. 26...Decoding circuit, Core...Output device rtS 3/...Predictive encoding circuit, 32...Run length or MH code 8' encoding circuit, 33
... Line memory, 3 Hiro... Adder, 3S... Reference point ROM % 36... Prediction RoM. 37...-Bedding disconnection circuit. Figure 1 (A) Figure 1 (B) Figure 3 Figure 7 (A) Figure 7 (B) A243 Figure 8 Figure 9 Figure 2 Figure 13 (A) Figure 13 (B) Figure 14 Figure 15 Figure 16 1 haiku of soil love

Claims (1)

[Claims] In an image data processing method that converts and encodes image data with gradations, the image data is converted to a value using a periodically arranged dither matrix as a comparison threshold, and the resulting value is output. For the above-mentioned Deizama)
1. A high-efficiency encoding method for image data, characterized in that reference pixels are determined according to the periodic arrangement of IJ blocks and predictive encoding is performed.